Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
JAAS
View Article Online
TUTORIAL REVIEW
Cite this: J. Anal. At. Spectrom., 2014,
29, 795
View Journal | View Issue
The potential for a carbon stable isotope biomarker
of dietary sugar intake
A. Hope Jahren,a Joshua N. Bostica and Brenda M. Davyb
Added sugar is sweetener added to foods during processing or preparation that offers no health benefits to
the consumer. The mean U.S. intake of added sugar is z16% of total calories; at the highest level of
consumption, this value exceeds 35%. In addition, 78% of added sugar typically consumed is refined from
C4 plants (e.g., corn and sugar cane) and it follows that the d13C value of these sweeteners is
conspicuously high compared to carbohydrates derived from C3 plants. We first suggested in 2006 the
potential for the d13C of human tissues to indicate corn- and cane-sugar intake for use in a clinical
setting. At present, self-reported dietary assessment methods are commonly used to measure added
sugar intake, but are subject to underreporting, particularly for sugar-rich foods. If a carbon isotope
technique could produce a quantitative indicator of dietary sugar intake, it would be invaluable to the
prevention and clinical treatment of chronic diseases associated with excess sugar consumption.
Research to date has focused upon testing the correlation between diet as characterized either by bulk
food d13C value or by food composition (e.g., added sugar intake quartile) and the d13C values of human
tissues. Analysis of hair, nail and red blood cells in modern humans with known diet has revealed
associations between the d13C value of bulk diet and the d13C value of these tissues. With respect to
added sugar, the d13C values of blood serum and fingerstick blood have both been shown to be
associated with added sugar intake, even after adjustment for meat intake. Researchers have attempted
to isolate specific compounds in blood that are uniquely derived from dietary carbohydrates, such as
direct endogenous carbohydrate sources (blood glucose) and specific non-essential amino acids (red
blood cell alanine), and have seen strong correlations with added sugar intake. Recognized dietary
Received 13th October 2013
Accepted 30th January 2014
confounders such as meat/animal products have been addressed using statistical adjustments and a
dual-isotope analytical approach that invokes d15N as a correction factor. Further controlled feeding
DOI: 10.1039/c3ja50339a
studies and epidemiological surveys complete with Institutional Review Board approval are needed to
www.rsc.org/jaas
establish the sensitivity of d13C tissue assay as an objective biomarker for added sugar intake.
a
University of Hawaii, School of Earth and Ocean Science and Technology, Honolulu,
HI, 96822, USA. E-mail: [email protected]
A. Hope Jahren, PhD, is a
Professor within the School of
Ocean and Earth Science and
Technology at the University of
Hawaii. She conducts laboratory
experiments on living plants and
analyzes plant fossils, with a
particular interest in the carbon
stable isotope composition of
plant tissues. She is a Fellow
and Medalist of both the
Geological Society of America
(GSA) and the American
Geophysical Union (AGU), and has authored more than 50 scientic publications.
This journal is © The Royal Society of Chemistry 2014
b
Virginia Tech, Department of Human Nutrition, Foods and Exercise, Blacksburg, VA,
24061, USA
Joshua Bostic is a research
technician in Hope Jahren's
laboratory. He oversees the lab's
nutrition program and manages
sample processing, data analysis, and methods development.
He received a BS in Human
Nutrition, Foods and Exercise
from Virginia Tech in 2012.
J. Anal. At. Spectrom., 2014, 29, 795–816 | 795
View Article Online
JAAS
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
1. Introduction
The stable isotope carbon 13 (13C) comprises more than 1% of
the carbon on Earth and its abundance varies by 10% across
terrestrial substrates. Since the 1970s, off-line isotope ratio
mass spectrometry (IRMS) techniques have existed for the
analysis of carbon stable isotopes in organic samples.1 The
coupling of continuous-ow techniques to IRMS has further
facilitated high-precision measurements of C, nitrogen (N),
oxygen (O), sulfur (S) and hydrogen (H) isotopes.2 Elemental
analyzers allow for the ash-combustion of solid organic
samples, while capillary gas and liquid chromatography interfaced to IRMS have made possible the compound-specic
isotope analysis of puried organic analytes. Modern plant,3
animal4 and fungal5 tissues have been extensively characterized
for carbon isotope composition, and the observed isotopic
fractionations between metabolic substrates and resultant
tissues have been shown to be inuenced by a myriad of factors,
including substrate availability,6 metabolic efficiency,7 growth
rate,8 trophic position,9,10 reproductive status,11 enzymatic
activity,12 energy availability,13,14 disease states,15,16 genetic
mutations17 and other processes internal and external to the
organism. Nonetheless, the d13C value of a living organism is
considered rst and foremost to reect the ultimate carbon
source for metabolism.18,19 Under this paradigm, numerous
studies have sought to characterize the isotopic composition of
the diet of free-living animals, including invertebrates,20 sh,21
reptiles,22 birds,23 mammals24 and primates.25 In contrast,
comparatively little work has been performed to establish the
use of carbon stable isotope analysis of clinically-sampled biological tissues in order to quantify the diet of “free-living”
humans. However, workers have performed extensive isotopic
analyses on the remains of fossil hominids in order to reconstruct diet during ancient26–28 and historical29 times. The interpretation of carbon isotopes in human fossil material usually
involves tracing the introduction of maize and/or millet into the
human diet, two grains with conspicuous carbon isotope
composition.30 The substrate of choice for most of these studies
was bone and tooth collagen,31 bone carbonate32 and tooth
Brenda Davy, PhD, RD, is an
Associate Professor in the
Department of Human Nutrition,
Foods and Exercise (HNFE) at
Virginia Tech. She conducts
applied and clinical nutrition
research addressing the inuence
of lifestyle and behavioral factors
on weight management and
health, with a particular interest
in the health effects of added
sugar consumption. She is a
Fellow of the American College of
Sports Medicine (FACSM) and The Obesity Society (FTOS), and has
authored more than 50 scientic publications.
796 | J. Anal. At. Spectrom., 2014, 29, 795–816
Tutorial Review
enamel33 due to the enhanced chemical preservation of these
substrates relative to other tissues aer burial.34 The isotopic
composition of mammalian diet is faithfully preserved in bone
and apatite35 and may be an ideal substrate for paleodiet
reconstruction, both because of resistance to diagenesis and
because these tissues, once constructed, are not remobilized for
metabolic processes. However, bone and tooth sampling is not
an option for clinical studies of living human subjects due to its
exceptional invasiveness. Researchers have opted to pursue
blood, ngernail, hair and other more expendable tissues
collected from living human subjects, and the interpretation of
such datasets is rapidly maturing. Our particular interest is the
potential for the d13C of human tissues to indicate corn- and
cane-sugar intake, as rst suggested in 2006.36 The goals of this
review are two-fold: rst, to comprehensively describe the
particular challenges unique to the role of sugar within human
metabolism, and second, to summarize current nutrition
research characterizing the relationship between diet and the
carbon isotope composition of human tissues within clinical
experiments. We note that although we rely upon statistics
characterizing the “American diet,” multiple studies have
shown that the recent dietary trends of many developed countries (particularly the United Kingdom and Germany) mirror
those of the United States.37
Carbon stable isotope technology has already been
extensively applied to medical science. Articial “spiking” of
carbon-compounds with very large (typically 3000 natural
abundance) amounts of 13C is commonly employed in
biomedical research (reviewed in ref. 38). The advantage of
natural abundance IRMS is that it does not require chemical
intervention, and instead uses the compound's natural 13Csignature as a chemical tracer. This allows for a variety of
compounds to be followed through the human body as the
result of ingestion without requiring a priori manipulation of
the compound. In addition, new continuous-ow interfaced
IRMS instruments lend themselves to rapid and highthroughput analyses39 and are no more costly to purchase
and operate than many analytical and diagnostic instruments found in hospital laboratories. Here we review issues
relevant to the development of an isotopic technique
that would employ the carbon isotope composition (d13C
value) of clinically-obtained human tissues as a quantitative
indicator of dietary sugar intake. If such an application were
possible, it would represent a major advance in the prevention and clinical treatment of chronic diseases associated
with excess sugar consumption.
2.
Sugar
Carbohydrates are the most abundant biomolecules on
Earth, and the oxidation of carbohydrates is the central
energy-yielding pathway in most heterotrophs. On average,
carbohydrates supply about 50% of total calories in the
American diet. In turn, about half of all carbohydrate intake
is in the form of simple sugars: monosaccharides and
disaccharides (Fig. 1).
This journal is © The Royal Society of Chemistry 2014
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
Tutorial Review
JAAS
Almost all carbohydrates are converted to glucose, the
primary source of cellular energy, during digestion and processing in the liver. Polysaccharides and disaccharides are
broken down into their monosaccharide subunits during
digestion, which are subsequently transported to the liver for
processing. A majority of glucose, which can be used by every
cell in the body, bypasses the liver and continues on to the
peripheral circulation. Galactose and fructose are primarily
taken up by the liver, where they are converted to glycolytic
intermediates to be used in the production of glucose or fatty
acid intermediates via gluconeogenesis or glycolysis, respectively. Glucose serves as the primary energy source immediately
aer feeding and performs many important functions in the
human body that fatty acids – the slower, more sustained source
of energy – cannot. The barrier separating the brain from the
rest of the body is impermeable to fatty acids; therefore, glucose
fuels the brain and central nervous system. Ketone bodies
produced from fatty acids can serve as a substitute fuel for the
brain, although their prolonged use may be detrimental.44,45 In
addition, glucose can be metabolized where oxygen delivery is
limited, such as within muscle during high intensity exercise,
while fatty acids cannot. Finally, the addition of glucose to
proteins and lipids via glycosylation is essential to the structure,
function, and regulation of numerous enzymes and cell
membrane components.46
2.2
Fig. 1 The majority of the edible sugars found in food are (a) glucose,
(b) fructose and (c) sucrose. High-fructose corn syrup (HFCS) is a mix
of glucose and fructose, meant to approximate the sweetness of
sucrose, which is an even combination of glucose and fructose; HFCS42 is 42% fructose and 58% glucose; HFCS-55 is 55% fructose and 45%
glucose.
2.1
Structure and function
Glucose (D-glucose is oen referred to as “dextrose”) is the hexose
monosaccharide that circulates within human blood as a ready
substrate for glycolytic metabolism. The three important dietary
disaccharides are maltose (two glucose units) formed from
partially-hydrolyzed starch, lactose (one glucose unit and one
galactose unit) found in milk, and sucrose (one glucose unit and
one fructose unit). Sucrose, a major intermediate product of
photosynthesis and the principal form of carbohydrate transported within plants, furnishes approximately one-third of the
carbohydrates within a typical diet. The recently-engineered dietary sugar high-fructose corn syrup (HFCS)40 may be the most
contentious component of the American diet. During the manufacture of HFCS, glucoamylase breaks various polysaccharides
from the corn plant into simple glucose units.41 The enzyme
glucose isomerase is then used to convert about half of the
molecules to fructose, resulting in a compound that approximates
the composition and sweetness of sucrose.42 Largely due to the
advent of HFCS, fructose per se currently comprises more than
one-third of all sugar in the American diet.43
This journal is © The Royal Society of Chemistry 2014
Dietary intake
Despite the importance of blood glucose to survival, dietary
sugars are not a necessary component of the human diet. The
Institute of Medicine has established a recommended daily
allowance (RDA) of 130 g per day for total carbohydrate,47 but it
need not be supplied by dietary sugars. All digestible dietary
carbohydrates (such as those found in fresh fruits and whole
grains) can be transformed into glucose, and along with
gluconeogenesis from non-carbohydrate sources, can provide
all of the glucose needed for survival.48 Nonetheless, almost half
of all carbohydrates consumed as part of the American diet are
delivered as the simple sugars listed in Table 1.49
Most of these sugars are rened and then added to foods in
the form of syrups during processing or preparation. The U.S.
intake of such added sugars (AS) has dramatically increased
during recent decades: the mean intake among U.S. adults is
currently equal to z16% of total calories (359 kcal per day), up
from 10.6% in the late 1970s.50,51 Current AS intake among
people $2 years of age is 83.9 g per day,52 equivalent to roughly
20 teaspoons of sugar per day. Among adults, the highest AS
intake levels are among men and women aged 19–50 years, with
approximately 70–80% of this population segment exceeding
the current AS intake recommendation found within the 2009
Scientic Statement by the American Heart Association.53,54
Elevated AS intake has been associated with a myriad of
negative health outcomes, including increased blood pressure,55 insulin resistance, fasting and postprandial dyslipidemia,56–59 and a low grade chronic inammatory state.60,61 The
results of large epidemiological trials have also linked AS intake
with increased risk of weight gain, obesity, type 2 diabetes, and
J. Anal. At. Spectrom., 2014, 29, 795–816 | 797
View Article Online
JAAS
Tutorial Review
Table 1
List of edible mono- and disaccharides and common food
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
sources
Sugar
Food Sources
Glucose
Fruits (esp. grapes), vegetables,
corn syrup, HFCS, honey, processed
meats, baked goods, sweetened beverages
Fruits (esp. apples, pears), vegetables,
HFCS, honey, frozen desserts,
sweetened beverages
Dairy products (as part of lactose; rarely
found in isolation)
High maltose corn syrup, hard candy,
frozen deserts, beer
Sugarcane, vegetables (esp. sugar beets),
fruits (esp. pineapple), baked goods,
breads & cereals, sweetened beverages
Dairy products
Fructose
Galactose
Maltose
(glucose + glucose)
Sucrose
(fructose + glucose)
Lactose
(glucose + galactose)
cardiovascular disease.62–65 Rened sugars offer no benecial
micronutrients or ber, as do natural sources of sugar, such as
fruits, vegetables, and dairy products. This is cause for concern,
as the overall nutritional quality of the diet is likely compromised with a high AS consumption.
Public health policy regarding AS consumption is crippled
by a lack of data available on whether reductions in AS intake
reduce the risk of cardiovascular disease (CVD) or obesity. The
American Heart Association has acknowledged that the
chronic effect of a high AS intake on CVD risk remains uncertain due to limited and inconclusive research.54 Although
calories from AS represent a large proportion of the dramatic
rise in caloric intake over the last 30 years,54 the role of AS and
sugar-sweetened beverages (SSB) in the development and
progression of obesity and related co-morbidities remains
controversial.66–73 Given the difficulties associated with quantifying the contribution of AS intake to rising obesity rates and
the paramount importance of determining such contributions,
the need for objective research methodologies in this area
remains substantial.
2.3
Urgent need for an objective biomarker
One particularly illustrative example is the crucial need for an
assessment of the adverse effects of AS intake on cardiovascular
health. Cardiovascular diseases (CVD) such as heart disease and
stroke are the leading cause of mortality in the U.S., and onefourth of cardiovascular deaths are estimated to be avoidable
through healthy lifestyle behaviors (e.g., dietary behaviors,
obesity prevention).74 The prevalence of CVD and its associated
costs are expected to increase in the coming years, making
primary prevention strategies critical to reducing the associated
health and economic burden. The relationship between AS
intake and risk of CVD, the leading cause of death among U.S.
adults,75 is a high-priority research area due to inconclusive
evidence regarding the adverse effects of usual AS intake, the
benets of recommended AS intake levels, and the mechanisms
by which AS intake inuences CVD development.
798 | J. Anal. At. Spectrom., 2014, 29, 795–816
Dietary reports of sugar intake often underestimate actual
intake, particularly for high-sugar consumers. Data from Bingham
et al.79 shows that the risk of obesity was significantly underestimated
by dietary recall of sucrose and fructose relative to estimates based on
a urinary biomarker.
Fig. 2
A common limitation of dietary studies is reliance upon selfreported dietary intake data, which is oen subject to a certain
degree of underreporting.72,76,77 This is a serious problem for
sugar-rich foods in particular, as they are the most likely to be
underreported by subjects.78 Fig. 2 illustrates a study in which
an objective biomarker of sugar intake (urinary sucrose/fructose) clearly showed that increased sugar intake was associated
with increased odds of obesity, while the exact opposite relationship appeared true when the analysis was performed using
self-reported (subjective) sugar intake data.79 The urgent need
for novel methods, including biomarkers, to objectively assess
dietary intake has been formally recognized by the Institute of
Medicine and the National Institutes of Health (i.e., PAR-12197).80–82
3.
Dietary assessment
Food intake records, dietary recalls, and food frequency questionnaires are the most commonly used self-reported dietary
assessment methods. Each of these methods has strengths and
limitations (Table 2).
3.1
Self-reported methods
Food intake records (i.e., diaries) require that study participants
record all food and beverage intake for 72–96 consecutive hours
following the instructions of a trained research technician. In
contrast, dietary recalls are the result of an interview: the
participant is asked to describe and estimate the portions of all
foods and beverages consumed over the most recent 24 hour
period. These interviews are performed unannounced and can
be collected in-person or via phone. Data from either method is
then analyzed using nutritional analysis soware, such as the
Nutrition Data System for Research (NDSR) developed by the
UMN (http://www.ncc.umn.edu/products/ndsr.html), and then
This journal is © The Royal Society of Chemistry 2014
View Article Online
Tutorial Review
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
Table 2
JAAS
Strengths and limitations of commonly-used self-reported dietary intake assessment methods
Method
Strengths
Limitations
Food intake record/diary
Less reliant on memory than recalls and FFQ
Includes detailed information on foods and
beverages consumed
High subject burden
Resource intensive – administered by trained
staff, analysis soware required
Recording error is common – missing foods,
inaccurate portion size estimation
Estimates usual intake over a short time frame
(days)
Literacy level can be an issue
Underreporting is common, eating behaviors
can be altered, forgotten foods
Multiple 24 hour recalls
Literacy level is less of an issue, compared to
records and FFQ
Includes detailed information on foods and
beverages consumed
Used in national nutrition surveillance
Recall procedure can be automated, computeradministered
Less likely to alter eating behaviors than
records, less error due to forgotten foods, if
standardized collection procedures and
prompts to probe for forgotten foods are
included
High subject burden
Resource intensive – administered by trained
staff, analysis soware required
Recording error is common – missing foods,
inaccurate portion size estimation
Relies on subject's memory
Estimates usual intake over a short time frame
(days, weeks)
Literacy level can be an issue
Underreporting and reporting bias could be
an issue
Food frequency questionnaire (FFQ)
Less resource-intensive than records or recalls
– does not need to be administered by trained
staff; self-administered
Estimates usual intake over a longer period of
time (months, years)
More feasibly used in large-scale trials than
records and recalls
Moderate subject burden
Must be checked for missing responses, which
are common
Literacy level is an issue
Less precise than records and recalls
Oen overestimates micronutrient intake,
underestimates kcal intake
Must be validated for use in the targeted study
population
used to estimate the participants' “usual” daily dietary intake.
Dietary recalls are generally considered superior to food intake
records, as individuals oen change their dietary intake when
asked to record it.50,83 Both methods are relatively costly, timeand resource-intensive, burdensome for respondents, and do
not provide information on longer-term, habitual intake
patterns (e.g., intake over months or years). In addition, these
methods are oen not feasible for use in large-scale investigations or in studies of low-literacy populations.84 Recently, the
National Cancer Institute has made available the web-based
Automated Self-Administered 24 hour Dietary Recall (ASA24)
system for researchers, which may overcome some of the limitations discussed above through the provision of a computerbased dietary recall and analysis process.
Food frequency questionnaires (FFQ) provide information
about usual dietary intake over a longer period (e.g., months,
years), and can focus either on general dietary habits (i.e., the
whole diet) or on the consumption of specic categories of
foods (e.g., beverages or fat intake). An example of a brief FFQ
providing information on beverage intake in the past 30 days,
including intake of sugar-sweetened beverages, is the Beverage
Intake Questionnaire (BEVQ-15).85 This FFQ takes approximately 3–4 minutes to complete, can be scored rapidly without
This journal is © The Royal Society of Chemistry 2014
soware, and is available to researchers at no cost. For a “whole
diet” FFQ, completion time for participants can range from a
few minutes to an hour, depending upon the number of foods
included on the questionnaire. FFQ may be more feasible than
food records or recalls for large-scale trials, but this method is
generally considered less precise, as it tends to overestimate
micronutrient consumption and underestimate calorie and
macronutrient consumption.84 In addition, FFQ must oen be
tailored for the population under study, as they include only a
limited list of foods and beverages; for example, an FFQ developed for the U.S. adult population may not be useful for a study
of adolescents. The universal problems with all self-reported
dietary assessment methods include underreporting, literacy
level, reliance on memory, commonly forgotten foods such as
beverages, and social desirability bias.86
3.2
Objective methods
Alternatives to using self-reported dietary intake methods
include the direct observation of food intake by a trained
observer, which is generally not feasible when studying freeliving populations. An additional objective measure is 24 h
urinary sucrose/fructose analysis, which assays the small
J. Anal. At. Spectrom., 2014, 29, 795–816 | 799
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
JAAS
amount of dietary sucrose and fructose that is normally absorbed intact in the small intestine and subsequently excreted by
the kidneys. This method represents a potential biomarker of
sugar consumption, and validation studies have been conducted
in both normal weight and obese adults.87,88 Urinary sucrose is
particularly indicative of the intake of AS,89 which may be due to
the quicker digestion and absorption of AS when compared to
natural sugar sources (e.g., fruit). We note, however, that urinary
sucrose/fructose only provides an indicator of dietary sugar intake
for the period during which the urine is collected and reects
both fructose consumed from AS and fructose consumed from
natural sources of sugar. Furthermore, this biomarker technique
requires that study participants collect all urine samples for two
to eight 24 hour periods, and that researchers verify the
completeness of study participant's urine collections by using
chemical indicators of compliance (e.g., para-aminobenzoic acid
or “PABA”). Because test subjects tend to view even one 24 h urine
collection as cumbersome, the urinary sucrose/fructose assay
may not be compatible with long-term or large-scale trials aimed
at determining relationships between AS intake and disease risk.
Validation studies (e.g., ref. 87) have determined that substantial
intra- and inter-individual variability exists with this technique,
even with a controlled diet including a standardized level of sugar
intake. Additionally, because the amount of sucrose absorbed is
controlled by the permeability of the small intestine, any diseases
or medications which alter gastrointestinal function and permeability (e.g., Celiac disease or the widely-used non-steroidal antiinammatory drugs (NSAIDs)) may interfere with the accuracy of
this approach to determine AS intake.90
4. Dietary biomarkers
A medical biomarker is a characteristic that can be objectively
measured and evaluated as an indicator of human biological
processes. Great advances in medical science have come from
the development of key biomarkers: one important example is
the measurement of the serum protein Troponin, a product of
heart muscle contraction, that is a sensitive indicator of recent
myocardial cell injury91 and is also predictive of 60 day mortality
and organ failure.92 Dietary biomarkers are different in that they
are biomarkers of exposure: they provide information on
previous dietary intake.78,93 Biomarkers can be described relative to the time frame during which dietary intake is reected,
with short-term biomarkers reecting intake over hours or days,
medium-term reecting intake over weeks or months, and longterm reecting intake over months or years.93,94 Dietary
biomarkers can also be classied as either recovery or concentration biomarkers.78,93 Recovery biomarkers measure the
excretion of the biomarker over a certain time frame; for
example, the nitrogen content of urine is a biomarker for dietary
protein intake.95 Concentration biomarkers are correlated with
intake, but do not directly measure actual dietary intake; for
example, the carotenoids found within fruits and vegetables96
can be measured within blood as a direct reection of intake.97
An ideal biomarker for AS intake would demonstrate validity
(i.e., measures what is intended), reliability (i.e., reproducibility), and the sensitivity to detect changes in AS intake over
800 | J. Anal. At. Spectrom., 2014, 29, 795–816
Tutorial Review
time in response to dietary manipulation or intervention.78,93,94
According to Kuhnle,78 an important but oen neglected issue
in developing biomarkers is validation. To validate a dietary
biomarker, actual dietary intake must be known; this is
accomplished with a controlled feeding study design. In addition to these criteria, the biomarker sample collection and
analysis should be possible at a reasonable cost for researchers,
the degree of invasiveness should be acceptable for study
participants, and the sample collection method should be
feasible for participants and researchers.76,93,94 The ideal
biomarker is non- or minimally-invasive, inexpensive and
specic.80 The measurement of the d13C value of human tissues
could meet many of these criteria, provided that the approach is
optimized aer validation.
4.1
Potential substrates for bioassay
Because carbon is the second most abundant element by mass
in the human body and is present in virtually every tissue,98
there are many potential substrates for the bioassay of d13C
value. In determining the applicability of specic tissues in a
biomarker setting, there are four key qualities to consider.
These include: (1) turnover rate: timescale of dietary history
reected by the biomarker (2) invasiveness: the amount of pain
or discomfort associated with sampling procedure (3) availability: amount of sample available in previously archived form
and (4) storage/preparation conditions: special storage conditions or processing steps required prior to analysis. Since all
human tissues can be found somewhere upon a continuum of
these traits, the selection of substrate for isotopic analysis
becomes a negotiation of multiple practical considerations.
While the isotopic analysis of a homogenized whole-body
has been performed upon small animals to yield an estimate of
dietary consumption,19 such sampling is obviously unattainable
for live human subjects. Once biomineralized, tooth enamel
does not reabsorb or reprecipitate,99 making it a potential
isotopic substrate free of turnover. However, the carbon in tooth
enamel reects only dietary intake during tooth formation,
which is usually completed between the rst 3–7 years of life.100
While carbon within tooth enamel is the preferred substrate for
isotopic analysis of deceased human tissues due to its resistance to post-mortem degradation,101 its high invasiveness and
low availability render it inappropriate for all but the most
specialized forensic or clinical studies.102 Collagen, the most
prevalent protein in bone, has been shown in controlled feeding
trials using mice to reect the isotope composition of the diet
with high precision between samples.19 The long turnover time
of bone collagen (e.g., years in adult long bones) makes it ideal
for determining the long-term feeding patterns of deceased
organisms.103 However, the extreme invasiveness associated
with sampling bone from living humans precludes its applicability in living human trials. Fingernail and hair sampling
presents less-invasive alternatives to bone sampling (composed
of the brous protein keratin,104 these tissues are relatively
resistant to microbial degradation105) and have been widely
adopted in both paleo- and modern dietary studies.106–109 The
sampling of these tissues is minimally invasive, can be
This journal is © The Royal Society of Chemistry 2014
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
Tutorial Review
performed without training, and subsequently requires no
special storage or shipping conditions aer collection.
However, assays of hair and nail tissue are complicated by
uncertainty as to the exact period of dietary history that their
growth represents: controlled studies involving sudden changes
in the isotopic composition of diet resulted in new steady-values
of d13C value in hair only aer 2.3 to 4.0 months.110,111 While hair
and ngernail present ideal substrates for determining dietary
patterns among populations with stable longer-term
dietary habits, limitations arise when considering shorter-term
dietary habits of populations consuming heterogeneous diets.
Substrates with relatively short turnover times may be obtained
by sampling carbon output pools, including urine and fecal
matter. Collection of these tissues is minimally invasive, and
the d13C value of urinary steroids has been validated as an antidoping test.112 Recently, urine and fecal d13C values were shown
to reect recent changes in red meat and sh intake in an 8 day
controlled feeding trial, while stable isotope values of whole
blood, a tissue with a slow turnover time, did not change aer
the dietary intervention period. Thus, urine and feces are
attractive substrates for tracking short-term dietary intake, but
sample collection, storage and handling nuisances have
hindered widespread adoption for biomarker studies.113
Blood represents the most widely sampled human tissue due
to the vast array of diagnostic information that can be obtained
from a single sample, as “every cell in the body leaves a record of
its physiological state in the products it sheds to the blood.”114
Massive archives of human blood exist worldwide, collected in
conjunction with a wealth of dietary and disease data. Examples
include the National Health and Nutrition Examination Survey
(NHANES), an ongoing study by the Centers for Disease Control
and Prevention (CDC) that combines interviews and physical
examinations,114 and the Atherosclerosis Risk in Communities
Study (ARIC) which has followed a cohort of 4000 individuals
aged 45–64 since 1987, collecting medical, social and demographic data.115 Blood can provide information about dietary
intake over a wide variety of timescales ranging from a few
hours to a few months depending on the fraction of blood that
is isolated for analysis.116,117 The disadvantages associated with
JAAS
using venous blood include the mildly invasive (although,
routine in clinical settings) nature of venipuncture, which
requires a trained phlebotomist for collection, and the need for
special storage and preparatory steps (e.g., centrifugation, lowtemperature transport and storage) prior to isotopic assay. One
promising alternative is the use of ngerstick blood, which is
non-invasive, can be performed with minimal equipment, and
can be stored and transported as a dry blot, mitigating many of
the disadvantages presented by venous blood.118 Many assays
essential to nutrition research have already been adapted to
ngerstick sampling (e.g., hemoglobin, glucose and cholesterol119) and ngerstick-based methods allowing for whole
genome transcript proling are currently in development.120
5.
Carbon isotopes in sugar
Plants are the only organisms that can make sugar out of
inorganic carbon; their metabolic synthetic processes (known
as “photosynthesis”) controls the carbon isotopic composition
of all sugars available in the human diet. Beets, grapes, sugar
cane and corn are some of the many plants from which sugar is
extracted, puried and subsequently used to sweeten other
foods. Photosynthesis acts to reduce carbon dioxide (CO2) from
the atmosphere and transform it into sugar via the Calvin cycle
(Fig. 3); this process discriminates against 13C, causing the d13C
value of plants tissues to be lower than the d13CO2 value available to the plant.121
The difference in carbon isotope value between a plant and
the atmosphere under which it has grown differs between plant
species by more than 20&, and is caused by processes both
internal (e.g., diffusion through the stomata) and external (e.g.,
the atmospheric concentration of CO2) to the plant (reviewed in
ref. 6). Isotopic fractionation during photosynthesis is most
strongly inuenced by the enzyme catalyzing the initial
carboxylation of CO2.122 The vast majority of plants employ the
enzyme ribulose biphosphate carboxylase-oxygenase (rubisco)
to x CO2 (Fig. 3a2) into 3-carbon phosphoglycerate (Fig. 3e),
which then enters the Calvin cycle to ultimately produce glucose
(Fig. 3f). This process, known as “C3 photosynthesis,” is the
Fig. 3 Intermediate carbon compounds involved in photosynthesis. “C3” photosynthesis employs the rubisco enzyme to fix carbon dioxide (a2)
as 3-phosphoglycerate (e) which then enters the Calvin cycle to ultimately produce glucose (f). During “C4” photosynthesis, carbon dioxide (a1) is
first fixed via the enzyme phosphoenolpyruvate carboxylase as oxaloacetate (b). Oxaloacetate (b) is then transformed to malate (c) via the
oxidation of NADPH, and transported across the membrane that separates the mesophyll cell from the bundle-sheath cell, where rubisco is
located. After transport, malate (c) is decarboxylated to produce pyruvate (d) and carbon dioxide (a2) via the reduction of NADP+, and this
regenerated carbon dioxide (a2) forms the substrate for rubisco-facilitated fixation.
This journal is © The Royal Society of Chemistry 2014
J. Anal. At. Spectrom., 2014, 29, 795–816 | 801
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
JAAS
Tutorial Review
carbon reduction pathway used by all citrus fruits, grapes,
berries, melons, apples, bananas and beets, and results in plant
tissue with d13C values ranging from approximately 24 to
32& (Fig. 4).
Out of the 400+ families of owering plants, less than 5%
contain species adapted for the pre-concentration of CO2 prior
to its xation by rubisco. This relatively rare pathway rst xes
carbon dioxide (Fig. 3a1) via the enzyme phosphoenolpyruvate
carboxylase (PEP carboxylase) as oxaloacetate (Fig. 3b), which is
then transformed to malate (Fig. 3c) and transported across a
membrane to where rubisco is located. Aer transport, malate
is decarboxylated to produce pyruvate (Fig. 3d). The resulting
CO2 (Fig. 3a2) serves as the substrate for rubisco-facilitated
xation. Because the rst product of xation is the four-carbon
compound oxaloacetate, this is known as “C4 photosynthesis”.
This pre-concentration allows the rubisco enzyme to work more
efficiently by excluding the interference of oxygen within the
bundle-sheath anatomy, leading to high net photosynthetic
rate.127 This mechanism also serves to greatly reduce the net
isotope fractionation imparted by photosynthesis128 and results
in plant tissue with d13C values ranging from approximately 11
to 19& (Fig. 4). These “C4 plants” represent only a small
number of the Earth's plants, and rarely approach 50% of the
species present even within the grassland ecosystems where
they are most commonly found.129 However, three important
sugar crops happen to be C4 plants: corn (Zea mays), sugar cane
(Saccharum officinale) and sorghum (Sorghum vulgare).
5.1
intrinsic sugars.131 Here, we further specify “added sugar” as
sweetener added to foods during processing or preparation
offering no health benets to the consumer. The primary
sources of AS in the U.S. diet include sugar-sweetened beverages
(43% of total AS), sugars and candy (16%), and sweetened grains
(e.g., cookies; 19%).50,51 Added sugars contribute a large portion
of the calories within the typical American diet: z13–16% of
total calories come from AS.50,51,132 Moreover, 78% of AS found in
the typical American diet were rened from C4 plants, e.g., corn,
cane syrups and sugars (Fig. 5).
The d13C value of a sweetener reects the isotopic composition of the plant from which it was rened;36 corn syrup, cane
sugar and molasses range in d13C value from 10 to 13&,
falling within the general range of C4 plants (Fig. 4). Many foods
high in AS content (e.g., candy) carry a C4-signature in their
carbon isotope composition (Table 3). Sugar sweetened beverages (SSB), which have been associated with weight gain,134–136
Sugar in food
A myriad of recent studies describe the negative health effects
of excess sugar consumption,130 yet many important sources of
nutrients, such as fruits and some vegetables, are sources of
Added sugars in the U.S. food supply, separated by type. Sugar
derived from C4 plants (corn and cane) form the majority of these
added sugars, totaling 78.1%. Data is for 2012, compiled by the United
States Department of Agriculture (USDA) Economic Research
Service.133
Fig. 5
List of processed foods that retain the C4-signature of the
carbon isotope composition of the sugars added (compiled from ref.
36 and previously unpublished data)
Table 3
Fig. 4 Carbon isotope composition of plants, food and human blood
serum, as compiled from large-scale studies. Data for 203 species of
C3 plants (ref. 123) is contrasted with data for 8 species of C4 plants
(ref. 124) that include all the major clades found in each group. The
d13C values of food products (both as ingredients and as sweetened
beverages) reflect the plant type of origin.36 The distribution of d13C
values within human serum125 spans the range of values attributed to
both C3 and C4 plants, and absolute value has been correlated to
sweetener intake at various levels of significance.126
802 | J. Anal. At. Spectrom., 2014, 29, 795–816
Food
d13C value [&]
“Runts” hard candy
“SweeTarts” hard candy
“Bottle caps” hard candy
York peppermint patty candy
Kellogg's corn pops cereal
Kellogg's frosted akes cereal
Nature's path sweetened corn akes cereal
Berry berry kix cereal
Chocolate cheerios cereal
General mills cocoa puffs cereal
General mills trix cereal
Heinz ketchup
Del Monte fruit cocktail syrup
Dried papaya (sweetened)
Dried pineapple (sweetened)
Breyer's cherry vanilla frozen dairy dessert
10.7
10.7
11.0
16.4
15.3
16.4
13.2
12.1
13.4
13.1
12.7
15.4
17.2
13.7
13.9
15.6
This journal is © The Royal Society of Chemistry 2014
View Article Online
Tutorial Review
JAAS
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
Table 4 Total calorie contribution of added sugar, separated by type.
Data from the USDA Economic Research Service Sugar Yearbook
Tables 2012–2013133
Sweetener
% of Total
calories consumed
High fructose corn syrup
Other corn syrups
Cane sugar
Beet sugar
Other sweeteners (e.g., honey, maple syrup)
Total added sugars
4.6
1.5
4.0
2.7
0.2
13.0
are a topic of intense controversy in terms of heath advice137 and
regulation through taxation.138 While beverages sweetened with
concentrated fruit juices have a d13C value similar to the likely
fruits of origin (e.g., orange, grape, apple), beverages sweetened
by cane sugar and/or HFCS (e.g., soda) have d13C values close to
10& (Fig. 4).
5.2
Sugar in the human body
Carbon isotope ratio mass spectrometry is a tool with exceptional power and precision. For illustration, within a sample
composed of 1,000,000 molecules of glucose, there are typically
11,125 13C12C5H12O6 molecules if the glucose was derived from
corn syrup. In contrast, if the glucose was derived from beet
sugar, it would typically contain 10,956 13C12C5H12O6 molecules. For context, modern continuous-ow IRMS techniques
allow for the resolution of <1 molecule per 1,000,000 in the
glucose sample above, making possible an abundantly clear
differentiation between corn and beet. In the units of isotopic
measure, the vast majority of C3 and C4 plants (and by extension, the sugars rened from these plants) are separated by 5 to
20 units permil [&]. The uncertainty associate with triplicate
measurements of d13C value on properly homogenized samples
rarely exceeds 0.1 unit permil [&].
The rened sugars that comprise AS represent a highlypalatable, minimally satiating source of calories that consumers
may nd pleasureable;136 consequently, AS consumption has
Simplified schematic depicting the incorporation of diet into human tissues. Arrows signify transformations; letters correspond to
descriptions presented in Table 5. Each component (e.g., protein, carbohydrate, fat, etc.) contains carbon that may be isotopically fractionated
during transformation, and eventual incorporation into human tissues such as red blood cells, fingernails and hair. In addition, carbon can be lost
due to use in oxidative phosphorylation at or near many of the transformations depicted.
Fig. 6
This journal is © The Royal Society of Chemistry 2014
J. Anal. At. Spectrom., 2014, 29, 795–816 | 803
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
JAAS
increased dramatically in the last several decades.139 Food sources of AS represent three of the top ten sources of calories in the
diet among Americans aged $2 years.52 In the U.S., HFCS intake
has increased by more than 1000% in the last thirty years,140 an
increase mirrored by trends in overweight and obesity.141,142 Table
4 shows the data breakdown of AS by type according to their total
calorie contribution. Extrapolating this data to total AS
consumption data, the average American consumer ingests
10.16% of his or her total calories in the form of corn- or canederived sugars and syrups,132,133 while this value exceeds 25% in
citizens with the highest levels of consumption.51
All of the above facts taken together promote great optimism
that a d13C-based biomarker for AS consumption can be developed for human tissues, with widespread potential impacts on
human health outcomes research. Because the carbon-isotope
composition of most AS is relatively conspicuous and can be
well-separated analytically from other sources of calories, it
stands to reason that consumption of AS might be highly
correlated with the d13C value of a tissue assay. However, the
myriad metabolic transformations that occur aer carbohydrates enter the digestive system greatly complicates the
hypothesis that isotopically, “we are what we eat.”143 Fig. 6
presents a schematic that illustrates the complex and contingent nature of carbon cycling in human tissues (see also Table
5). Though the schematic is highly simplied, it effectively
highlights the liver as a clearing house and re-packaging center
for proteins, carbohydrates and fats, mainly in the form of
amino acids, glucose/glycogen and fatty acids/triglycerides.
Once a carbohydrate (e.g., AS) is digested and absorbed into the
bloodstream as a monosaccharide such as glucose (Fig. 6A), it
may be immediately used for energy by any cell in the body via
glycolysis and aerobic respiration or combined with an amino
group via transamination to form a non-essential amino acid
(Fig. 6N), the building blocks of tissue protein (Fig. 6L). This
glucose may also be used to synthesize glycogen, the storage
form of carbohydrate, via glycogenesis (Fig. 6C), or converted to
fat via de novo lipogenesis (Fig. 6E and G). Glycogen and fat,
which are stored in a variety of tissues, can be broken down and
used for energy production or amino acid synthesis when
carbon input is low (i.e., aer a meal). The above suggests that
carbohydrate in the diet could eventually be incorporated into
most types of tissues available for assay, including proteins (e.g.,
hair, ngernails, red blood cells), carbohydrates (e.g., blood
glucose), or fats (e.g., blood lipids). However, each of the
processes illustrated by Fig. 6 and listed in Table 5 may impart
an isotopic fractionation based on preferential incorporation of
12
C over 13C (or vice versa) and the magnitude of fractionation
specic to each process is currently unknown, though they may
be approximated when integrated across the whole-organism.144
This may be further confounded by the fact that each of the
tissues available for assay listed above may also contain
substantial carbon ultimately derived from ingested proteins
and fats. Finally, it is important to remember that much of what
we consume is oxidized for energy, rather than incorporated
into tissues during growth or maintenance. In fact, most of the
carbon consumed as food is ultimately exhaled as CO2 aer use
in oxidative phosphorylation at or near many of the
804 | J. Anal. At. Spectrom., 2014, 29, 795–816
Tutorial Review
Processes that transform diet within the human body;
designations correspond to the lettered arrows within Fig. 6
Table 5
Designation
Process
Description
A
Digestion
B
Transport
C
D
E
Glycogenesis
Glycogenolysis
De novo
lipogenesis
F
G
Glyceroneogenesis
Triglyceride
synthesis
H
Lipolysis
I
Gluconeogenesis
J
Lipoprotein
synthesis
K
Deposition
L
Protein synthesis/
degradation
M
Urea cycle
N
Amino acid
synthesis
Carbohydrates, fats and proteins
ingested from food and drink are
broken down within the mouth,
stomach and intestines into
(primarily) monosaccharides,
fatty-acids and amino-acids,
respectively
Transport across biological
membranes
Glucose is converted to glycogen
Glycogen is mobilized as glucose
Glucose is broken down into
acetyl CoA for fatty acid synthesis
and subsequent triglyceride
synthesis
Glucose is converted into glycerol
Fatty acids, both free and within
soluble lipoproteins, are
converted into triglycerides
Triglycerides are converted into
glycerol and free fatty acids
Glycerol and/or amino acids are
converted to glucose
Endogenous and exogenous lipids
are incorporated with amino acids
to form water-soluble lipoproteins
Triglycerides are concentrated
into adipose tissue or muscle
Synthesis/degradation of complex
tissues such as muscle, nails, hair
and/or blood cells from
constituent amino acids
Aer amino acids are
enzymatically deaminated, the
residual amino group is converted
to urea and excreted
Glucose is combined with a
nitrogen source to synthesize nonessential amino acids
transformations depicted in Fig. 6. For these reasons, the
literature to date has focused upon testing the correlation
between diet as characterized either by bulk d13C value or by
food composition (e.g., AS intake quartile) and the d13C values of
human tissues.
6. Carbon isotope studies
6.1
Interpreting biomarker statistics
The ultimate goal of nutritional biomarkers is to accurately
predict consumption of a certain food/nutrient. Statistical
approaches aimed at establishing the correlation between a
proposed biomarker and food/nutrient intake are oen utilized
in such investigations, with simple correlational and linear
regression analyses being common examples. In the case of a
stable isotope biomarker of AS intake, simple correlational or
linear regression analyses are used to assess associations
This journal is © The Royal Society of Chemistry 2014
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
Tutorial Review
between tissue d13C value and the amount of dietary AS
consumed, which is provided by the correlation coefficient (r).
Linear regression can also be used to determine if signicant
correlations persist aer controlling for possible confounding
variables, as described later (see Section 8.1). For d13C, linear
regression can be used to determine if the proposed biomarker
is independently associated with AS intake, aer accounting for
the potential dietary confounds of non-sweetener corn and
animal product consumption.145 Accounting for potential confounding variables in statistical analyses may also improve the
validity of biomarker as a predictor of dietary intake.145
Correlation coefficient interpretation is somewhat subjective. Thus, some researchers opt to provide the coefficient of
determination (R2) instead. The coefficient of determination,
which is simply the correlation coefficient squared, provides the
amount-percent of the variation in the dependent variable (AS
intake) that can be explained by variation in the independent
variable (stable isotope biomarker) (Taylor). P-Values are
provided with r and R2 values, in order to denote the level of
signicance of the associations between variables. Correlations
measured between nutritional biomarkers and reported dietary
intake vary widely in research trials, with typical r values
ranging from 0.03–0.70, with a mean of 0.39.118,146 While these
correlations may be interpreted by some as weak to modest,
they may underestimate the true validity of these biomarkers
due to the inherent inaccuracy of self-reported dietary intake
measures, and r values of 0.5–0.7 are typically considered
acceptable as a “realistic degree” of precision in dietary validation studies.146 Determining the validity of a biomarker in a
clinical/research setting also requires consideration of a
nutrient's diagnostic importance and the accuracy of current
dietary assessment methods in determining intake. Hence,
biomarkers with weaker relationships to a specic nutrient may
still be considered for clinical application if the nutrient of
interest is of high clinical importance and is oen misreported
in conventional dietary assessment methods.
Given the myriad factors involved in biomarker data interpretation, assigning “strengths” to the r and R2 values obtained
during analysis is somewhat subjective, and the strengths
ascribed to correlations will differ between studies and across
disciplines. To avoid confusion in the following sections and
provide coherency, correlations reported in the following
sections will be assigned strengths based on the following scale:
0.1–0.3 ¼ weak/small; 0.4–0.6 ¼ moderate; 0.7–0.9 ¼ strong.147
6.2
JAAS
hair as an isotopic substrate for large-scale epidemiological and
clinical studies where ease of sampling and storage conditions
are oen decisive factors in substrate selection. Early studies in
contemporary humans compared the isotope values of human
hair with whole diet. The rst of these studies, conducted by
Nakamura et al.,148 compared the d13C values of common food
items from Chicago, Tokyo, and Munich to the d13C values of
hair from human subjects living in each city. The study showed
a positive correlation between the d13C values of human hair
and the d13C values of dietary protein. In keeping with the
earliest experiments on animals,144 Nakamura et al. also found
that human hair was enriched in 13C by 1.7& relative to whole
diet.148 Similar enrichment values of 1.4& were observed by
Schoeller et al.,149 who compared the d13C values of common
food items to the d13C values of hair samples from 40 individuals living in Chicago. In contrast, Yoshinaga et al. discovered
larger and more variable carbon isotope enrichments (1.8 to
4.8&) between the diet and hair of indigenous populations
residing in undeveloped areas, which was possibly the function
of seasonal variations in native dietary food sources.151 Schoeller
et al.149 suggested that the d13C value of human hair was
correlated with the consumption of C4 plants. This relationship
was later tested by Huelsemann et al.,111 who conducted a 4
week controlled dietary change in which subjects went from
consuming a traditional German diet (61% of dietary carbon
from C3 plants and 31% of dietary carbon from terrestrial
animal meat) to a diet based on C4 plants (71% of dietary
carbon) and marine animal products (8% of dietary carbon),
with no change in total calories consumed. The C4 plant foods,
which comprised the bulk of the experimental diet, consisted of
corn, millet, amaranth, and cane sugar minimally supplemented by C3 vegetables in order to meet nutrient requirements. As expected, all subjects demonstrated signicant
increases in hair d13C values of 1.7 to 3.9& over the course of
Human hair
Hair is the most widely-used substrate for isotopic analysis as a
method
of
dietary
reconstruction
in
modern
humans.106,107,111,148–160 Several studies have investigated the use
of stable isotopes in human hair as a biomarker of food
consumption patterns through controlled feeding trials.
Human hair is a complex substrate composed of a variety of
protein and lipid components. The bulk of a hair's sha is
composed of a-keratin, a brous protein that imparts rigidity
and stability. This stability, combined with the relatively
noninvasive nature of hair sampling techniques, recommends
This journal is © The Royal Society of Chemistry 2014
Fig. 7 Change in carbon isotope composition in male (left) and female
(right) hair in response to dietary change. Open circles represent the
d13C value of diet prior to (“pre-exp” 1.5&) and during (“exp” + 0.7&)
the 28 day feeding experiment. Closed circles represent the d13C value
of hair prior to feeding (“pre-exp”) and the maximum value seen in hair
5.5 months after the beginning of the dietary change (“exp”). Data from
Tables 3 & 4 of Huelsemann et al.111
J. Anal. At. Spectrom., 2014, 29, 795–816 | 805
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
JAAS
Tutorial Review
the feeding trial (Fig. 7), but did not reach a new steady state
during the four-week experiment. Based upon the above, the
d13C value of hair may be an effective biomarker of mediumterm dietary intake (weeks/months), but data on sensitivity to
AS intake per se is at present lacking.
6.3
Human ngernails
Nail tissue is in many ways similar to hair tissue: ngernail and
toenail are noninvasive substrates primarily composed of the
protein keratin. Both human hair and nails grow at a relatively
constant rate among individuals, allowing for a rough timescale
to be determined based on the placement of the sample. As
might be predicted from the above, hair and nail tissues from
the same person were found to be similar with respect to d13C
value, with hair keratin being slightly enriched (+0.21&) in d13C
value compared to nail keratin.153 The rst study to apply nail
isotope analysis in a nutritional biomarker setting was conducted by Nardoto et al.,109 who sought to determine if the
isotopic composition of ngernail allowed for differentiation
according to diet in 816 individuals living in both urban and
rural regions of the United States and Brazil (Fig. 8).
Guided by FFQ-data on types of foods consumed and
frequency of animal protein consumption, the study determined the d13C value of foods commonly consumed in each
region. They observed the ngernails of urban Brazilian
subjects to be enriched in d13C value by +3.9& relative to those
of rural Brazilian subjects; they also observed the ngernails of
omnivore subjects to be enriched in d13C value by +1.5& relative
to those of vegetarian subjects within the urban Brazilian population. Finally, they observed the ngernails of urban Brazilian
subjects to be enriched in d13C value by +3.4& relative to those
of urban U.S. subjects. These isotopic differences were reected
in the isotopic signature of food samples from each region, as
meat products from Brazil were enriched in d13C value by +6.7&
relative to similar products from the U.S. In a similar study, the
Fig. 8 Carbon isotope composition of fingernail samples from people
living in Europe, Brazil and the United States. Open circles represent
individuals living in isolated areas subsisting on small-scale fishing and
agriculture; “error bars” represent the standard deviation of each
sample population (n: 6 to 455). Within the urban populations of both
Brazil and the U.S., vegetarian subjects tended to exhibit d13C values
that were z1& lower than that of omnivorous subjects. Data from
Table 3 of Nardoto et al.109
806 | J. Anal. At. Spectrom., 2014, 29, 795–816
authors showed a signicant enrichment of +0.9& in nail tissue
from the highest socioeconomic classes compared to the lowest
class in an urban Brazilian population, which was attributed to
higher consumption of processed foods and so drinks
amongst the affluent in urban Brazil.161
The signicant delay between ngernail formation and
sampling present logistical issues which may limit the applicability of nail as a biomarker substrate. The growth rate of
ngernails is 100 mm per day;162 therefore, a 1 mm length of
nail reects approximately ten days of dietary history. However,
nails grow from the root outward, and an average ngernail
length represents about six months of growth,162 although this
is highly variable based on age, race, and sex.163 Given the large
displacement and variability between nail synthesis and
sampling availability, nail substrates are most appropriate for
studies seeking to determine population-wide dietary intake
trends among subjects with consistent dietary patterns, and
could thus be considered reective of long-term dietary intake.
As with hair, data on sensitivity to AS intake per se has not been
gathered.
6.4
Red blood cells
Red Blood Cells (RBC; erythrocytes), White Blood Cells (WBC;
leukcocytes), and many other specic cell-types may be isolated
from whole blood164 and analyzed for isotopic composition.
RBCs, which comprise approximately 45% of total blood
volume, transport oxygen and carbon dioxide to and from the
lungs, and have an average lifespan of 120 days,165 suggesting
that RBCs may incorporate dietary intake across 3–4 months.
Nash et al.166 measured associations between RBC d13C values
and intakes of “traditional” versus “market” foods in a Native
Alaskan population, and found a moderate positive correlation
(r ¼ 0.46, p < 0.0001) between market food consumption and
RBC d13C. The authors noted that “market” food included a
variety of products that directly and indirectly incorporated C4
plant content, including corn and sugarcane-based products as
well as meat from corn-fed animals, which likely strengthened
their ndings compared to similar studies specifying added
sugars as the dependent variable. In the authors' follow-up
study, which broke down the category of “market foods” into
total sugar, commercial meats, and corn products, RBC d13C
exhibited a greater association with commercial meat intake
than total sugar intake. While the d13C value of RBC offers a
compelling biomarker for determining intake of total C4-based
foods, the greater association with protein-based food items
limits its use in isolation in an AS biomarker application,
although dual-isotope and specic-compound approaches may
increase the specicity of RBC analysis for sugar-based food
items (discussed below).
6.5
Human whole blood and blood serum
The hair, nail and red blood cell studies discussed above were
largely designed to explore the potential for the d13C value in
tissues to reect the d13C value of whole diet, with special
attention to dietary protein. Indeed, most of the carbon within
these substrates is in the form of various proteins, and the
This journal is © The Royal Society of Chemistry 2014
View Article Online
Tutorial Review
JAAS
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
Table 6 Carbon isotope composition of blood serum from a
randomly-chosen population of 406 individuals; data from ref. 125
d13C
[&] dataset
Male
n ¼ 201
Female
n ¼ 205
Total
n ¼ 406
Mean
Maximum
Minimum
SD
Range
19.0
16.3
21.2
0.8
4.9
19.3
15.8
23.2
0.8
7.4
19.1
15.8
23.2
0.8
7.4
Table 8 Group characteristics of low and high AS consumers; data
from ref. 118
Characteristics
Male : female (n : n)
Age (mean SEM)
BMIc (mean SEM)
Fingerstick blood
d13C value (mean SEM)
a
b
direct incorporation of digested amino acids into such tissues
via protein synthesis (Fig. 6L) is a straightforward model for
biomarker interpretation. However, substrates derived from
dietary carbohydrates are also converted to non-essential amino
acids and incorporated into proteins. These carbohydratederived amino acids represent a signicant proportion of the
total amino acid pool. The carbon in whole blood is dominated
by proteins (e.g. albumin, immunoglobulins, hemoglobin).
Nonetheless, blood is an extremely attractive substrate because
it is universally collected and archived; the American Red Cross
and the American Association of Blood Banks alone collect and
process more than 7 million liters of blood each year.167 Whole
blood may be cheaply and quickly processed to exclude certain
proteins before archival: aer collection whole blood is oen
separated into fractions via centrifugation, which promotes a
clot largely composed of RBC and WBC. The supernatant,
known as “plasma” if the clotting factors (brinogen) are
retained and “serum” if they are removed, is largely composed
of the proteins albumin and globulin. If the conspicuously high
d13C value of most AS relative to other carbohydrates were
observable within the blood collected from subjects with high
AS intakes, a widely-applicable biomarker of great clinical and
public health value might be developed.
Kra et al.125 represents the largest-scale study of stable
isotopes in human blood, which reported the d13C value of 406
anonymous subjects with unknown dietary habits (e.g., potentially non-fasting) in order to quantify the isotopic variability in
a free-living population (Table 6).
The total range in the d13C value of both clot and serum was
observed to span the range in d13C value seen across C3 and C4
plant-derived sweeteners (Fig. 4). No enrichment in d13C value
was observed between clot and serum. Population histograms
Mean serum d13C value stratified by SSB consumption in the
ARIC study; data from ref. 126
Table 7
Sweetened beverage
consumption
Serum
d13C [&]
#1 per week
2–4 per week
$5 per week
19.47a
19.35
19.15a
a 13
d C value was signicantly lower for those consuming sweetened
beverages infrequently (#1 per week) compared with those
consuming such beverages at a high frequency ($5 per week, P ¼
0.004 by t-test of difference).
This journal is © The Royal Society of Chemistry 2014
Low SSB
consumptiona,b
(98 16 kcal)
High SSB
consumptiona,b
(236 63 kcal)
8 : 12
47.8 4.1 years
26.4 0.9 kg m2
20.06 0.17&
10 : 10
38.1 3.8 years
29.1 2.5 kg m2
19.59 0.13&
Determined by Beverage Intake Questionnaire (BEVQ; ref. 168).
Mean SEM (standard error of the mean). c Body mass index.
for d13C values in blood revealed normal distributions for both
males and females, with mean values for both close to 19&.
The study also showed no isotopic difference between capillary
blood and venous blood, obtained from 24 fasting individuals.
Finally, the study conrmed that common processing techniques such as the addition of blood additives (i.e., sodium
uoride and polymerized acrylamide resin), sample handling
(i.e., autoclaving and freeze drying) and refrigerated storage up
to 115 days do not alter the original d13C value of blood samples.
To date, two studies have evaluated the association between
AS intake in the form of sugar-sweetened beverages (SSB) and
the d13C value of whole blood. Yeung et al.126 conducted a crosssectional study within the ARIC trial of 186 adults, comparing
FFQ data to serum d13C value (Table 7).
Mean serum d13C value was signicantly lower for those
consuming SSB infrequently (#1 per week) compared with
those consuming such beverages at a high frequency ($5 per
week), even aer statistical adjustment for questionnairereported animal fat and “whole corn” (e.g., tortillas) consumption. A second trial, conducted by Davy et al.,118 employed the
non-invasive ngerstick sampling technique in order to evaluate the association between the d13C value of whole blood and
AS intake (Table 8). The study found that the d13C value of ngerstick blood was associated (p < 0.05) with both AS intake and
total SSB intake. These ndings suggest that the d13C value of
human blood, which is commonly collected in many clinical
and epidemiological studies, is a promising biomarker for AS
and SSB intake, with the time frame of dietary exposure reected likely being medium-term (weeks/months).
7. Specific-compound approaches
The studies discussed directly above imply that, while the
conspicuously high d13C value of AS may be observable within
the protein-dense portions of blood as it is commonly collected
and processed (e.g., centrifuged serum fraction, ngerstick
blot), differences in AS consumption can only be demonstrated
at present between very high- and low-end consumers.118,126 An
ideal biomarker of AS would permit the sensitive determination
of AS intake across a continuum of consumption levels, allowing for the establishment of correlations between AS intake,
health outcomes and/or chronic disease risk. Towards this, new
J. Anal. At. Spectrom., 2014, 29, 795–816 | 807
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
JAAS
Tutorial Review
studies have attempted to isolate specic compounds in blood
that are more likely to be derived from dietary carbohydrates,
such as direct endogenous carbohydrate sources and specic
non-essential amino acids.
7.1
Specic compound: blood glucose
Direct measurement of glucose in the blood is an attractive
potential substrate for an isotope-based biomarker of AS intake,
as it is an obvious and important destination for dietary
carbohydrate. In addition, blood glucose concentration is a
routinely monitored index of health, a myriad of methods exist
for measurement,169 and blood-glucose dynamics have been
exhaustively studied.170 Blood glucose originates from three
sources: dietary glucose which bypasses the liver (Fig. 6A),
glycogenolysis of glycogen stored in the muscle and liver tissue
(Fig. 6D), or certain non-carbohydrate precursors via gluconeogenesis (Fig. 6I). Gluconeogenesis may be of primary
importance only during long-term fasting, as experiments using
13
C-labeled amino acids indicated that non-carbohydrates
sources did not constitute a major source of blood glucose aer
an overnight fast.171 Recent measurements comparing the d13C
value of serum glucose (isolated via liquid chromatography) to
the bulk d13C value of the same serum sample revealed a large
amount of unexplained variation between the substrates (R2 ¼
0.52; Fig. 9), which supports the heterogeneous origin of the
various blood components.
Cook et al.116 specically tested the potential of plasma
glucose as a biomarker of AS consumption: in their crossover
study, subjects consumed 3 weight-maintaining diets with
varying amount of carbohydrates derived from cane sugar and
HFCS (5, 16, and 32% of total carbohydrates per day). Each diet
lasted for 7 days, and all meals were provided to the participants; fasting blood was drawn in the morning, and non-fasting
blood was drawn in the aernoon on days 1–6 and at 2 hour
intervals throughout day 7. Signicant differences were not
detected between the d13C values of plasma glucose in fasting
subjects, regardless of diet. However, strong positive correlations were observed between the C4-sugar content of breakfast
and the d13C value of post-breakfast (10 a.m.) blood glucose (R2
¼ 0.89) and between total daily C4-sugar content of diet and the
d13C value of aernoon blood glucose on days 1–6 (R2 ¼ 0.90).
Based on a regression of their data against pure-C3 to pure-C4
carbon end-members, Cook et al.116 contended that almost all of
the carbon in blood glucose was derived from dietary carbohydrate sources immediately aer a meal, while only two-thirds of
blood glucose sampled later in the day was derived from
carbohydrates. These claims agree with prior metabolic tracer
studies demonstrating the gradual shi from use of carbohydrates in the meal to the use of liver glycogen and gluconeogenesis to maintain blood glucose levels.173,174 The lack of
signicant differences between the d13C value of fasting plasma
glucose in subjects consuming different diets is somewhat
surprising, as dietary carbohydrates would be expected to
contribute to liver glycogen stores during the fed state (Fig. 6C),
followed by release of these stores during the overnight fast
(Fig. 6D). In contrast, it appears that a majority of blood glucose
in these fasted individuals was derived from non-carbohydrate
sources via gluconeogenesis (Fig. 6I). The relatively small
concentration of glucose in blood (z0.8 mg mL1) and the
immense complexity and individual variability in blood glucose
homeostasis175 must be considered when interpreting isotope
results in relation to the contribution of gluconeogenesis in the
fasted state. Serum glucose d13C value may be the best
biomarker of acute sugar consumption (i.e., a recent meal);
however, dietary recall or direct observation are non-invasive
methods commonly used for such purposes.
7.2
Fig. 9 Correlation between the bulk d13C value of serum and the d13C
value of serum glucose as isolated via liquid-chromatography IRMS;
“error bars” reflect the standard deviation seen across 3–6 replicate
measurements (Jahren unpublished data). Subjects were 30 middle
aged and older U.S. adults enrolled in a water-weight loss study.172 The
lack of correlation between the two datasets suggests a heterogeneous origin for the carbon within different blood components.
808 | J. Anal. At. Spectrom., 2014, 29, 795–816
Specic compound: RBC alanine
Non-essential amino acids (NEAAs) may be synthesized from
the intermediates of carbohydrate metabolism in a variety of
tissues. Certain NEAAs, particularly alanine, are more likely to
be derived from dietary carbohydrates. The higher proportion of
blood alanine derived from carbohydrate is due to its role in the
glucose–alanine cycle in which pyruvate, the product of anaerobic glucose metabolism, is transaminated to form alanine.
This alanine is subsequently returned to the liver via the
bloodstream, where the carbon skeleton can be used to produce
new glucose via gluconeogenesis (Fig. 6A).176,177 Based on this,
Choy et al.117 hypothesized that the d13C value of alanine isolated from RBC might be a more sensitive biomarker of AS
intake than the d13C value of bulk RBC.178 Utilizing RBC samples
from a dual-isotope trial,145 researchers measured the carbon
isotope composition of NEAAs likely to be derived from glucose
(alanine, glycine, serine, proline, aspartate/asparagine, and
glutamate/glutamine) via gas chromatography IRMS to determine associations between NEAA d13C values and dietary intake
data gathered via four 24 h dietary recalls spread over 4 weeks.
Of the amino acids tested, only alanine (CH3CH(NH2)COOH)
was signicantly correlated (r ¼ 0.59, p < 0.0001) with AS intake,
This journal is © The Royal Society of Chemistry 2014
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
Tutorial Review
and particularly with SSB intake (r ¼ 0.70, p < 0.0001). Moreover,
alanine d13C value was not associated with any form of animal
protein intake, including commercial meat. While RBC alanine
is a promising biomarker of AS intake, alanine may be derived
from two sources: from the combination of glucose and a
nitrogen source within the liver (Fig. 6N) or from the direct
digestion of animal (e.g., meat, seafood, eggs, gelatin) and plant
(e.g., beans, nuts, soy, corn, rice) proteins (Fig. 6A). The pathway
most used will depend upon the nitrogen balance and carbohydrate intake of the individual: states of negative nitrogen
balance (i.e., low protein intake or starvation) and/or high
carbohydrate intake will result in a greater proportion of nonessential amino acids derived from carbohydrates compared to
states of positive nitrogen balance and/or low carbohydrate
intake.179,180 While most Americans consume adequate levels of
dietary protein to remain in nitrogen balance equilibrium,181
the continuum of carbohydrate and protein consumption may
produce inter- and intra-individual differences in biomarker
sensitivity to SSB intake. Thus, while RBC alanine is a promising
candidate biomarker of SSB consumption, further testing in a
more heterogenous population with a broad range of carbohydrate and protein intake is need to determine its applicability to
the general population.
7.3
Specic compound: hemoglobin A1c
The ideal biomarker for AS intake is sensitive to dietary carbohydrate intake but not susceptible to acute uctuations due to
meal-to-meal, or day-to-day changes in dietary intake; it would
also be a long-lived bodily substance that is primary derived
from ingested carbohydrates. Hemoglobin A1c analysis may
meet all of these requirements. Hemoglobin, which comprises
90% of the mass of RBC, is the protein that permits the intertissue transport of oxygen. Hemoglobin A1c (“Hb A1c”; also
knows as “glycated hemoglobin”) is hemoglobin with a glucose
molecule covalently bound to the N-terminal valine of the beta
chain. This bound glucose serves no role in the function of
hemoglobin; however, a useful relationship between glycated
hemoglobins and diabetes was rst recognized in 1971.182 Later
studies established that the formation of Hb A1c occured
through non-enzymatic “glycation” at a slow constant rate that
was directly proportional to the concentration of glucose in
blood.183 Because the average lifespan of hemoglobin is about
12 weeks, determination of Hb A1c/total Hb ratio is the gold
standard of long-term blood glucose control in the clinical
setting.184 An advantage of Hb A1c over traditional blood
glucose measurements as a potential isotopic biomarker is its
relative unresponsiveness to recent feeding status or to acute
changes in blood glucose concentration. Because the concentration of Hb A1c slowly increases within an RBC over the course
of its lifespan, measurement of Hb A1c concentration allows for
the calculation of the average blood glucose concentration over
3 month period.185 The same advantages that Hb A1c confers
to the assessment of blood glucose concentration should
theoretically apply to the measurement of the isotopic composition of blood glucose, i.e., Hb A1c may hold the key to determining the average isotopic composition of dietary
This journal is © The Royal Society of Chemistry 2014
JAAS
carbohydrates over a long period of time, provided the associated analytical challenges can be overcome.
The glucose moiety makes up only a very small proportion of
the total carbon present in Hb A1c; therefore, isolation and
concentration of glucose would be necessary prior to the
determination of d13C value. The recently-discovered enzyme
Fructosyl Amino Acid Oxidase (FAOD) can be used to break the
glycolytic bond in Hb A1c, resulting in the products glucosone
(C6H10O6), valine (HO2CCH(NH2)CH(CH3)20), and hydrogen
peroxide (H2O2).186 The FAOD method has recently been applied
to Hb A1c analysis through the colorimetric determination of
hydrogen peroxide production, and many commercial test kits
are now available.187,188 In developing a protocol for the isolation
and carbon-isotope analysis of glucose derived from Hb A1c,
modications to protocols currently used in commercial FAOD
kits will be required, as they indirectly measure Hb A1c via an
assay of the hydrogen peroxide product. Because the product of
interest for d13C analysis is glucosone, free blood glucose must
be removed from the sample prior to the FAOD reaction step in
order to minimize interference. Furthermore, because FAOD
cannot react with intact Hb A1c, a proteolytic digestion step is
required to liberate glycated amino acids.186 FAOD can then be
introduced to break the glycolytic bonds, and the glucosone
product may be isolated via liquid or gas chromatography for
subsequent d13C analysis. Provided that isolation and isotopic
analysis of the bound glucose is successful, Hb A1c could offer a
potential biomarker of long-term AS consumption that warrants
validation in clinical trials.
8. The carbon isotope composition of
meat as a confounding variable
Corn and cane sugar, and thus a large proportion of AS (Fig. 5),
gure prominently in the typical American diet; however, they
are not on the only dietary sources which possess conspicuously
high d13C values. The U.S. typically produces 64 billion dollars
worth of Zea mays each year, and corn is the primary feed
supplied to milk and meat animals.189 The d13C value of cornbased silage and meal ranges from 14 to 11&,190 in close
agreement with corn-based foods available to humans (Table 3).
Somewhat ironically, Americans may even consume beef that
has been fed upon foods extremely high in AS and particularly
in HFCS: in 2012 it was widely-reported that drought-induced
corn shortages rendered it cheaper for feedlot operators and
dairy producers to feed “co-products” consisting of hard and
so candies (e.g., marshmallows).191
Meat products reect the isotopic composition of the feed
supplied to the animal via the same mechanisms implicated
within the earliest dietary isotope studies on laboratory
animals.19 As a result, an analysis of the d13C value in meat is
commonly used to authenticate claims of “free ranging” and/or
“grass-fed” animal protein.192–195 Widely-consumed and cheaply
available sources of meat are likely to have a corn-fed carbon
isotopic signature, as shown in the analysis of meat from more
than 300 hamburgers (Fig. 10) (mean d13C value ¼ 18.0&;
n ¼ 162) and chicken sandwiches (mean d13C value ¼ 17.5&;
J. Anal. At. Spectrom., 2014, 29, 795–816 | 809
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
JAAS
Fig. 10 Distribution in d13C values of beef sampled from 161
hamburgers purchased at fast-food restaurants throughout the
continental United States (ref. 196) According to the values for muscle
and lipid published in Bahar et al.,193 beef with d13C > 19.5& implied a
“final diet of corn silage”; 71% of the burgers sampled met this
particular criterion.
n ¼ 161) from fast-food distributors across the continental
United States.196 In addition, fried convenience foods are oen
cooked in corn oil197 that retains the d13C value of the plant from
which it has been extracted.198 It follows from the above, that
any heterogeneous human tissue sampled (e.g., whole blood)
may have been synthesized from carbon with d13C value ¼ 18
to 11& (the range associated with most AS) that was not
contributed to the diet as carbohydrate, but was instead
contributed as meat or associated fat. This may be particularly
true for tissues comprised mainly of amino acids such as hair,
nail, RBCs, and whole blood (Fig. 6L).
The potential of meat consumption to confound any C4carbohydrate carbon isotope signal within human tissues has
been recognized; a positive correlation between d13C values in
human tissues and d13C values of dietary protein has been
reported by several studies.106,107,148,149 Nonetheless, the association between the d13C value of human blood and AS intake has
persisted in serum126 even aer adjustment for meat intake.
Considered practically, the continuous ow system that
converts carbon to CO2 also converts nitrogen to N2, which are
then separated prior to entering the mass spectrometer within a
ow of helium.199,200 Thus the isotopic analysis of both carbon
and nitrogen on a single sample is possible given the system's
ability to reliably switch tuning parameters within a short time
frame. Partially because of this, many (if not most) of the carbon
isotope studies discussed above also reported the d15N value of
substrates analyzed. Nash et al.145 rst formally suggested the
use the nitrogen stable isotope composition (i.e., 15N/14N
expressed as d15N [&]) of human tissues in order to produce a
“correction factor” for animal protein consumption and
increase biomarker sensitivity for carbohydrate intake. Such an
approach is based on the long-standing observation that 15N is
concentrated with each increase in trophic level.201 The exact
mechanism for the concentration of 15N with trophic level
remains uncertain, but may be related to absolute nitrogen
intake in a mechanism akin to bioaccumulation.202 Based on this
810 | J. Anal. At. Spectrom., 2014, 29, 795–816
Tutorial Review
principle, d15N has been used to delineate the ecological niche of
species,203 as well to determine the relative proportion of plant
versus animal protein in ancient human diets.204 Petzke et al.155
presented d13C and d15N data from the hair of individuals selfreported as omnivores (n ¼ 99), ovo-lacto vegetarians (n ¼ 15) and
vegans (n ¼ 6), and showed that the d15N value of omnivores was
enriched by z+2& relative to ovo-lacto vegetarians, which in
turn was enriched by z+1.5& relative to vegans. The total ranges
in both the d13C values and the d15N values of blood serum in the
large population reported by Kra et al.125 were observed to span
the ranges in d13C and d15N values established by Petzke et al.155
for omnivores and ovo-lacto vegetarians (Fig. 11).
Nash et al.'s166 assessment of traditional versus market food
intake within a Native Alaskan population revealed a moderate
association between the d15N value of RBCs and intake of
aquatic animals (r ¼ 0.52). Based upon this association, this
group sought to increase the validity of RBC d13C value as a
biomarker of SSB intake through the use of RBC d15N value to
control for marine food intake.145 This dual isotope model
notably improved the amount of variation in AS consumption
that was explained by d13C value (R2 ¼ 0.33), when compared to
using d13C value alone (R2 ¼ 0.03). Using this dual isotope
model to estimate total added sugar intake in a larger Yup'ik
population, this group found positive correlations between AS
intake and several markers of metabolic syndrome, demonstrating the potential utility of a stable-isotope biomarker in
clinical trials aimed at assessing the effects of AS intake on longterm health.205 We note, however, that d15N values were not
associated with commercial meat intake in the original model.
This was likely due to the proportionally large intake of marine
animals relative to commercial meats in the study population,
as marine animals are more enriched in d15N value than their
terrestrial counterparts.206 In addition, the d15N value of
Fig. 11 The stable carbon and nitrogen isotope composition of human
hair (closed circles) and blood serum (open circle); “error bars” reflect
the standard deviation associated with n individuals tested. Hair
samples were contributed by self-identified omnivores and ovo-lactovegetarians in Germany (ref. 155); blood serum was contributed by 101
male and 105 female anonymous patients at the Johns Hopkins
Medical Institutions (ref. 125).
This journal is © The Royal Society of Chemistry 2014
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
Tutorial Review
terrestrial meat animals is variable: Huelsemann et al.207 reported
that beef, pork and chicken varied by more than 3.0& across a
dataset of twenty-two samples. Furthermore, tissue d15N values
may be affected by other physiological conditions, including
caloric availability, total protein content of the diet, and pathological conditions such as liver disease.180 Thus, further testing in
the general U.S. population is warranted to determine the relationship between tissue d15N and commercial meat intake and
validate the dual isotope model as a strategy for increasing the
sensitivity of d13C-based biomarkers.
9. Special considerations:
institutional review board approval
Developing solutions to many of today's societal challenges,
particularly those in the realm of public health,208 requires the
collaboration of scientists from different disciplines who may
not have worked together historically. The research area
addressed by this review encompasses elds of geochemistry,
nutritional science, food science, physiology, metabolism and
biochemistry; collaborators with expertise in epidemiology and
statistics may also be needed. Interdisciplinary research teams
have the potential to achieve scientic innovations and
advances which are not possible to achieve working as a lone
investigator in a single discipline,208,209 although the challenges
facing interdisciplinary research teams can be signicant. One
critical example is the Institutional Review Board (IRB) approval
process required before any study involving interventions or
interactions with living human subjects can be performed.
Although only about half of all English-language biomedical
journals require proof of IRB-approval prior to publication,210
articles may be formally retracted if IRB-approval cannot be
veried. For example, the British Journal of Anaesthesia states
categorically that “Lack of IRB approval means that the research
was unethical” within its retraction policy.211 This requirement
exists for any sampling of human tissue for analysis within a
research or teaching setting, no matter how informal. Most
research institutions exert punitive measures against
researchers found to be in violation of IRB, including a prohibition of supervising or advising research. Of particular concern
are informal experiments associated with curriculum: if
students feel that their grade in a course is contingent upon
providing a tissue sample for analysis, this constitutes “undue
pressure to enroll” which most institutions view as a serious
ethics violation (e.g., Johns Hopkins IRB Student Recruitment
Policy).212 Each major research institution has its own IRB
comprised of a committee of volunteers including scientists,
non-scientists, community members, and health care professionals. The Board's mission is to ensure research protocols
involving human subjects are ethical and that the rights of
participants are protected. Prior to initiating any human
research studies, the IRB will review all study documents,
including the protocol, informed consent document, forms or
questionnaires used for data collection, and recruitment
advertisements. The IRB insures that human subjects are fully
informed of all risks and benets of participating in the study,
This journal is © The Royal Society of Chemistry 2014
JAAS
and provides signicant protection to the researcher once the
results of the study come under scrutiny for publication. In
addition, the study investigators and research staff performing
the types of isotope experiments described above are likely
required to have certied training in the protection of human
subjects and in the handling of biohazards. For studies
involving blood this might be obvious, but to a scientist unfamiliar with human subject research this requirement could
pose a signicant barrier to initiating research with human
subjects. Prior to the initiation of any work, investigators are
urged to contact their IRB office, and learn the specics of what
their institution requires.
10. Conclusions
There is an urgent need for new techniques to directly evaluate
the effects of added sugars on human health. Global sugar
consumption is increasing rapidly, particularly with respect to
fructose (e.g., as high fructose corn syrup), which now
comprises more than one-third of all sugar in the American
diet.43 Adverse outcomes ranging from major depression213 to
breast cancer214,215 to antisocial behavior216 have been postulated to be related to added sugar consumption. At present,
there are no denitive objective biomarkers of medium- or longterm added sugar intake in clinical use. Because of this urgency,
even moderate correlations between the d13C value of a given
tissue and dietary added sugar intake make the tissue in
question a valid substrate for further exploration. Despite the
myriad of metabolic transformations that preclude a simple
relationship between the isotope composition of carbohydrates
that enter the digestive system and the isotope composition of
human tissues available for assay, early clinical data exploiting
the conspicuous d13C value associated with corn- and canederived added sugar describes a highly promising potential
biomarker of added sugar intake. Continued research into
alternative tissue substrates and analytical processes aimed at
mitigating interference by confounding variables is warranted
to increase the applicability of a carbon stable isotope
biomarker in large clinical and epidemiological trials.
Acknowledgements
This work is dedicated to the late Christopher Dyer Saudek,
M.D., founder and director of the Johns Hopkins Comprehensive Diabetes Center. We are grateful to have benetted from his
encouragement, and continue to be inspired by his memory.
Notes and references
1 J. Frazer, Mikrochim. Acta, 1962, 50, 993–999.
2 J. T. Brenna, T. N. Corso, H. J. Tobias and R. J. Caimi, Mass
Spectrom. Rev., 1997, 16, 227–258.
3 L. A. Cernusak, N. Ubierna, K. Winter, J. A. Holtum,
J. D. Marshall and G. D. Farquhar, New Phytol., 2013, 950–
965.
4 K. A. Hobson, Oecologia, 1999, 120, 314–326.
J. Anal. At. Spectrom., 2014, 29, 795–816 | 811
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
JAAS
5 P. Högberg, A. H. Plamboeck, A. F. S. Taylor and
P. M. A. Fransson, Proc. Natl. Acad. Sci. U. S. A., 1999, 96,
8534–8539.
6 B. A. Schubert and A. H. Jahren, Geochim. Cosmochim. Acta,
2012, 96, 29–43.
7 E. Brugnoli, K. T. Hubick, S. von Caemmerer, S. C. Wong
and G. D. Farquhar, Plant Physiol., 1988, 88, 1418–
1424.
8 M. R. Henn, G. Gleixner and I. H. Chapela, Appl. Environ.
Microbiol., 2002, 68, 4956–4964.
9 J. H. McCutchan, W. M. Lewis, C. Kendall and
C. C. McGrath, Oikos, 2003, 102, 378–390.
10 J. F. Kelly, Can. J. Zool., 2000, 78, 1–27.
11 S. D. Newsome, P. L. Koch, M. A. Etnier and D. AuriolesGamboa, Mar. Mammal Sci., 2006, 22, 556–572.
12 G. Tcherkez and G. D. Farquhar, Funct. Plant Biol., 2005, 32,
277–291.
13 D. E. Butz, S. L. Casperson and L. D. Whigham, J. Anal. At.
Spectrom., 2014, Advance article.
14 L. D. Whigham, D. E. Butz, L. K. Johnson, D. A. Schoeller,
D. H. Abbott, W. P. Porter and M. E. Cook, Int. J. Obes.,
2014, Epub ahead of print.
15 J. P. Boriosi, D. G. Maki, R. A. Yngsdal-Krenz, E. R. Wald,
W. P. Porter, M. E. Cook and D. E. Butz, J. Anal. At.
Spectrom., 2014, Advance article.
16 D. E. Butz, S. L. Morello, J. Sand, G. N. Holland and
M. E. Cook, J. Anal. At. Spectrom., 2014, Advance article.
17 J.-P. Godin, A. B. Ross, M. Cléroux, E. Pouteau, I. Montoliu,
M. Moser and S. Kochhar, PLoS One, 2013, 8, e74866.
18 N. C. Arens, A. Hope Jahren and R. Amundson, Paleobiology,
2000, 26, 137–164.
19 M. J. DeNiro and S. Epstein, Geochim. Cosmochim. Acta,
1978, 42, 495–506.
20 O. Schmidt, J. P. Curry, J. Dyckmans, E. Rota and
C. M. Scrimgeour, Pedobiologia, 2004, 48, 171–180.
21 B. Gu, C. L. Schelske and M. V. Hoyer, J. Fish Biol., 1996, 49,
1233–1243.
22 U. Struck, A. V. Altenbach, M. Gaulke and F. Glaw,
Naturwissenschaen, 2002, 89, 470–473.
23 K. A. Hobson and R. G. Clark, Condor, 1992, 94, 189–197.
24 M. Ben-David, R. W. Flynn and D. M. Schell, Oecologia,
1997, 111, 280–291.
25 M. J. Schoeninger, J. Moore and J. M. Sept, Am. J. Primatol.,
1999, 49, 297–314.
26 J. Lee-Thorp, J. F. Thackeray and N. van der Merwe, J. Hum.
Evol., 2000, 39, 565–576.
27 R. G. Klein, Proc. Natl. Acad. Sci. U. S. A., 2013, 110, 10470–
10472.
28 M. Sponheimer, Z. Alemseged, T. E. Cerling, F. E. Grine,
W. H. Kimbel, M. G. Leakey, J. A. Lee-Thorp, F. K. Manthi,
K. E. Reed, B. A. Wood and J. G. Wynn, Proc. Natl. Acad.
Sci. U. S. A., 2013, 110, 10513–10518.
29 H. Bocherens, M. Mashkour, D. G. Drucker, I. Moussa and
D. Billiou, J. Archaeol. Sci., 2006, 33, 253–264.
30 H. P. Schwarcz and M. J. Schoeninger, Am. J. Phys.
Anthropol., 1991, 34, 283–321.
31 S. H. Ambrose, J. Archaeol. Sci., 1990, 17, 431–451.
812 | J. Anal. At. Spectrom., 2014, 29, 795–816
Tutorial Review
32 A. M. Katzenberg, Advances in stable isotope analysis of
prehistoric bones, Wiley-Liss, New York, 1992.
33 M. Balasse, International Journal of Osteoarchaeology, 2002,
12, 155–165.
34 Y. Wang, T. E. Cerling and B. J. MacFadden, Palaeogeogr.,
Palaeoclimatol., Palaeoecol., 1994, 107, 269–279.
35 S. Jim, S. H. Ambrose and R. P. Evershed, Geochim.
Cosmochim. Acta, 2004, 68, 61–72.
36 A. H. Jahren, C. Saudek, E. H. Yeung, W. H. Kao, R. A. Kra
and B. Caballero, Am. J. Clin. Nutr., 2006, 84, 1380–1384.
37 S. N. Heller and L. R. Hackler, Am. J. Clin. Nutr., 1978, 31,
1510–1514.
38 F. Stellaard and H. Elzinga, Isot. Environ. Health Stud., 2005,
41, 345–361.
39 B. Fry, W. Brand, F. J. Mersch, K. Tholke and R. Garritt,
Anal. Chem., 1992, 64, 288–291.
40 L. M. Hanover and J. S. White, Am. J. Clin. Nutr., 1993, 58,
724S–732S.
41 S. Chiba, Biosci., Biotechnol., Biochem., 1997, 61, 1233–1239.
42 S. H. Bhosale, M. B. Rao and V. V. Deshpande, Microbiol.
Rev., 1996, 60, 280–300.
43 J. P. Bantle, S. K. Raatz, W. Thomas and A. Georgopoulos,
Am. J. Clin. Nutr., 2000, 72, 1128–1134.
44 G. F. Cahill, Jr, Annu. Rev. Nutr., 2006, 26, 1–22.
45 O. E. Owen, A. P. Morgan, H. G. Kemp, J. M. Sullivan,
M. G. Herrera and G. F. Cahill, Jr, J. Clin. Invest., 1967, 46,
1589–1595.
46 R. G. Spiro, Glycobiology, 2002, 12, 43R–56R.
47 Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat,
Fatty Acids, Cholesterol, Protein, and Amino Acids
(Macronutrients), The National Academies Press, 2005.
48 A. Sparti, H. Milon, V. Di Vetta, P. Schneiter, L. Tappy,
E. Jequier and Y. Schutz, Am. J. Clin. Nutr., 2000, 72,
1461–1468.
49 A. K. Kant, Am. J. Clin. Nutr., 2000, 72, 929–936.
50 R. K. Johnson and B. A. Yon, J. Am. Diet. Assoc., 2010, 110,
1296–1299.
51 B. P. Marriott, L. Olsho, L. Hadden and P. Connor, Crit. Rev.
Food Sci. Nutr., 2010, 50, 228–258.
52 P. J. Huth, V. L. Fulgoni, 3rd, D. R. Keast, K. Park and
N. Auestad, Nutr. J., 2013, 12, 116.
53 S. M. Krebs-Smith, P. M. Guenther, A. F. Subar,
S. I. Kirkpatrick and K. W. Dodd, J. Nutr., 2010, 140,
1832–1838.
54 R. K. Johnson, L. J. Appel, M. Brands, B. V. Howard,
M. Lefevre, R. H. Lustig, F. Sacks, L. M. Steffen and
J. Wylie-Rosett, P. A. American Heart Association
Nutrition Committee of the Council on Nutrition,
Metabolism, E. the Council on and Prevention,
Circulation, 2009, 120, 1011–1020.
55 I. J. Brown, J. Stamler, L. Van Horn, C. E. Robertson,
Q. Chan, A. R. Dyer, C. C. Huang, B. L. Rodriguez,
L. Zhao, M. L. Daviglus, H. Ueshima and P. Elliott, M. M.
International Study of and G. Blood Pressure Research,
Hypertension, 2011, 57, 695–701.
56 K. L. Stanhope, J. M. Schwarz, N. L. Keim, S. C. Griffen,
A. A. Bremer, J. L. Graham, B. Hatcher, C. L. Cox,
This journal is © The Royal Society of Chemistry 2014
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
Tutorial Review
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
A. Dyachenko, W. Zhang, J. P. McGahan, A. Seibert,
R. M. Krauss, S. Chiu, E. J. Schaefer, M. Ai, S. Otokozawa,
K. Nakajima, T. Nakano, C. Beysen, M. K. Hellerstein,
L. Berglund and P. J. Havel, J. Clin. Invest., 2009, 119,
1322–1334.
K. L. Teff, J. Grudziak, R. R. Townsend, T. N. Dunn,
R. W. Grant, S. H. Adams, N. L. Keim, B. P. Cummings,
K. L. Stanhope and P. J. Havel, J. Clin. Endocrinol. Metab.,
2009, 94, 1562–1569.
K. L. Stanhope, A. A. Bremer, V. Medici, K. Nakajima, Y. Ito,
T. Nakano, G. Chen, T. H. Fong, V. Lee, R. I. Menorca,
N. L. Keim and P. J. Havel, J. Clin. Endocrinol. Metab.,
2011, 96, E1596–E1605.
M. Maersk, A. Belza, H. Stodkilde-Jorgensen, S. Ringgaard,
E. Chabanova, H. Thomsen, S. B. Pedersen, A. Astrup and
B. Richelsen, Am. J. Clin. Nutr., 2012, 95, 283–289.
I. Aeberli, P. A. Gerber, M. Hochuli, S. Kohler, S. R. Haile,
I. Gouni-Berthold, H. K. Berthold, G. A. Spinas and
K. Berneis, Am. J. Clin. Nutr., 2011, 94, 479–485.
C. L. Cox, K. L. Stanhope, J. M. Schwarz, J. L. Graham,
B. Hatcher, S. C. Griffen, A. A. Bremer, L. Berglund,
J. P. McGahan, N. L. Keim and P. J. Havel, J. Clin.
Endocrinol. Metab., 2011, 96, E2034–E2038.
F. B. Hu and V. S. Malik, Physiol. Behav., 2010, 100, 47–54.
V. S. Malik, B. M. Popkin, G. A. Bray, J. P. Despres and
F. B. Hu, Circulation, 2010, 121, 1356–1364.
L. de Koning, V. S. Malik, M. D. Kellogg, E. B. Rimm,
W. C. Willett and F. B. Hu, Circulation, 2012, 125, 1735–1741.
V. S. Malik, B. M. Popkin, G. A. Bray, J. P. Despres,
W. C. Willett and F. B. Hu, Diabetes Care, 2010, 33, 2477–
2483.
D. B. Allison and R. D. Mattes, JAMA, J. Am. Med. Assoc.,
2009, 301, 318–320.
K. D. Brownell, T. Farley, W. C. Willett, B. M. Popkin,
F. J. Chaloupka, J. W. Thompson and D. S. Ludwig, N.
Engl. J. Med., 2009, 361, 1599–1605.
V. S. Malik and F. B. Hu, Am. J. Clin. Nutr., 2011, 94, 1161–
1162.
D. L. Weed, M. D. Althuis and P. J. Mink, Am. J. Clin. Nutr.,
2011, 94, 1340–1347.
M. D. Huffman, Circulation, 2012, 125, 1718–1720.
J. Slavin, Nutr. Rev., 2012, 70(suppl. 2), S111–S120.
R. C. Cottrell, BMJ, 2013, 346.
W. C. Willett and D. S. Ludwig, BMJ, 2013, 346, e8077.
L. J. Schieb, S. A. Greer, M. D. Ritchey, M. G. George and
M. L. Casper, Morbidity and Mortality Weekly Report, 2013,
62, p. 721–727.
S. L. Murphy, L. Xu and K. D. Kochanek, Deaths: Final Data
for 2010, CDC National Vital Statistics Reports, 2013.
F. E. Thompson, A. F. Subar, C. M. Loria, J. L. Reedy and
T. Baranowski, J. Am. Diet. Assoc., 2010, 110, 48–51.
E. C. Kosova, P. Auinger and A. A. Bremer, J. Acad. Nutr.
Diet., 2013, 113, 219–227.
G. G. Kuhnle, J. Sci. Food Agric., 2012, 92, 1145–1149.
S. Bingham, R. Luben, A. Welch, N. Tasevska, N. Wareham
and K. T. Khaw, Cancer Epidemiol., Biomarkers Prev., 2007,
16, 1651–1654.
This journal is © The Royal Society of Chemistry 2014
JAAS
80 C. W. Suitor and L. D. Meyers, Dietary Reference Intakes
Research Synthesis: Workshop Summary, The National
Academies Press, 2006.
81 D. S. Hardin, J. Am. Diet. Assoc., 2009, 109, 1698–1699.
82 B. McCabe-Sellers, J. Am. Diet. Assoc., 2010, 110, 52–54.
83 F. E. Thompson and A. F. Subar, in Nutrition in the
Prevention and Treatment of Disease, Academic Press, 3rd
edn, 2013, pp. 5–46.
84 W. C. Willett and E. Lenart, in Nutritional Epidemiology,
O.U. Press, New York, 2nd edn, 1998, pp. 101–147.
85 V. E. Hedrick, J. Savla, D. L. Comber, K. D. Flack,
P. A. Estabrooks, P. A. Nsiah-Kumi, S. Ortmeier and
B. M. Davy, J. Acad. Nutr. Diet., 2012, 112, 840–849.
86 R. K. Johnson, B. A. Yon and J. H. Hankin, in Research:
Successful Approaches, ed. E. R. Monsen and L. Van Horn,
American Dietetic Association, 3rd edn, pp. 187–204.
87 N. Tasevska, S. A. Runswick, A. McTaggart and
S. A. Bingham, Cancer Epidemiol., Biomarkers Prev., 2005,
14, 1287–1294.
88 A. M. Joosen, G. G. Kuhnle, S. A. Runswick and
S. A. Bingham, Int. J. Obes., 2008, 32, 1736–1740.
89 N. Tasevska, S. A. Runswick, A. A. Welch, A. McTaggart and
S. A. Bingham, Eur. J. Clin. Nutr., 2009, 63, 653–659.
90 H. Kawabata, J. B. Meddings, Y. Uchida, K. Matsuda,
K. Sasahara and M. Nishioka, J. Gastroenterol. Hepatol.,
1998, 13, 1002–1006.
91 C. W. Hamm, J. Ravkilde, W. Gerhardt, P. Jorgensen,
E. Peheim, L. Ljungdahl, B. Goldmann and H. A. Katus,
N. Engl. J. Med., 1992, 327, 146–150.
92 E. K. Bajwa, P. D. Boyce, J. L. Januzzi, M. N. Gong,
B. T. Thompson and D. C. Christiani, Crit. Care Med.,
2007, 35, 2484–2490.
93 A. J. Cross and J. W. Lampe, in Nutrition in the prevention
and treatment of disease, ed. A. Coulston, C. Boushey and
M. Ferruzzi, Academic Press, Waltham, MA, 3rd edn, 2013.
94 V. E. Hedrick, A. M. Dietrich, P. A. Estabrooks, J. Savla,
E. Serrano and B. M. Davy, Nutr. J., 2012, 11, 109.
95 S. A. Bingham, J. Nutr., 2003, 133(suppl. 3), 921S–924S.
96 A. R. Mangels, J. M. Holden, G. R. Beecher, M. R. Forman
and E. Lanza, J. Am. Diet. Assoc., 1993, 93, 284–296.
97 K. J. Yeum, S. L. Booth, J. A. Sadowski, C. Liu, G. Tang,
N. I. Krinsky and R. M. Russell, Am. J. Clin. Nutr., 1996,
64, 594–602.
98 Z. Wang and R. N. Pierson, Jr, Phys. Med. Biol., 2010, 55,
5953–5963.
99 H. C. W. Skinner and A. H. Jahren, in Biogeochemistry,
ed. W. H. Schlesinger, Elsevier Science, 2003, vol. 8,
pp. 117–184.
100 D. J. Reid and M. C. Dean, J. Hum. Evol., 2006, 50, 329–
346.
101 B. T. Fuller, M. P. Richards and S. A. Mays, J. Archaeol. Sci.,
2003, 30, 1673–1684.
102 K. Pye, Geol. Soc. Spec. Publ., 2004, 232, 215–236.
103 S. Ambrose and L. Norr, in Prehistoric Human Bone, ed.
J. Lambert and G. Grupe, Springer, Berlin Heidelberg,
1993, pp. 1–37.
104 H. C. Hopps, Sci. Total Environ., 1977, 7, 71–89.
J. Anal. At. Spectrom., 2014, 29, 795–816 | 813
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
JAAS
105 G. E. Mathison, Ann. Soc. Belg. Med. Trop. Parasitol. Mycol.,
1964, 44, 767–791.
106 K. J. Petzke, B. T. Fuller and C. C. Metges, Curr. Opin. Clin.
Nutr. Metab. Care, 2010, 13, 532–540.
107 K. J. Petzke, H. Boeing, S. Klaus and C. C. Metges, J. Nutr.,
2005, 135, 1515–1520.
108 S. A. Macko, M. H. Engel, V. Andrusevich, G. Lubec,
T. C. O'Connell and R. E. Hedges, Philos. Trans. R. Soc., B,
1999, 354, 65–75, discussion 75–66.
109 G. B. Nardoto, S. Silva, C. Kendall, J. R. Ehleringer,
L. A. Chesson, E. S. Ferraz, M. Z. Moreira, J. P. Ometto and
L. A. Martinelli, Am. J. Phys. Anthropol., 2006, 131, 137–146.
110 J. S. McCullagh, J. A. Tripp and R. E. Hedges, Rapid
Commun. Mass Spectrom., 2005, 19, 3227–3231.
111 F. Huelsemann, U. Flenker, K. Koehler and W. Schaenzer,
Rapid Commun. Mass Spectrom., 2009, 23, 2448–2454.
112 A. T. Cawley and U. Flenker, J. Mass Spectrom., 2008, 43,
854–864.
113 G. G. Kuhnle, A. M. Joosen, C. J. Kneale and T. C. O'Connell,
Eur. J. Nutr., 2013, 52, 389–395.
114 L. A. Liotta, M. Ferrari and E. Petricoin, Nature, 2003, 425,
905.
115 Atherosclerosis Risk in Communities Study Description, ed.
UNC Department of Biostatistics, 2013, vol. 2013.
116 C. M. Cook, A. L. Alvig, Y. Q. Liu and D. A. Schoeller, J. Nutr.,
2010, 140, 333–337.
117 K. Choy, S. H. Nash, A. R. Kristal, S. Hopkins, B. B. Boyer
and D. M. O'Brien, J. Nutr., 2013, 878–884.
118 B. M. Davy, A. H. Jahren, V. E. Hedrick and D. L. Comber,
J. Am. Diet. Assoc., 2011, 111, 874–878.
119 B. Sterling, T. Kiang, K. Subramanian, M. Saltman,
W. Smart, M. Tsay, J. Sugarman, D. Patel, D. Monger,
D. Martin, et al., Clin. Chem., 1992, 38, 1658–1664.
120 E. H. Robison, T. S. Mondala, A. R. Williams, S. R. Head,
D. R. Salomon and S. M. Kurian, BMC Genomics, 2009, 10,
617.
121 R. Park and S. Epstein, Geochim. Cosmochim. Acta, 1960, 21,
110–126.
122 G. D. Farquhar, J. R. Ehleringer and K. T. Hubick, Annu. Rev.
Plant Physiol. Plant Mol. Biol., 1989, 40, 503–537.
123 A. Hope Jahren, Review of Palaeobotany and Palynology,
2004, 132, 291–313.
124 R. Amundson, R. R. Evett, A. H. Jahren and J. Bartolome,
Review of Palaeobotany and Palynology, 1997, 99, 17–24.
125 R. A. Kra, A. H. Jahren and C. D. Saudek, Rapid Commun.
Mass Spectrom., 2008, 22, 3683–3692.
126 E. H. Yeung, C. D. Saudek, A. H. Jahren, W. H. Kao, M. Islas,
R. Kra, J. Coresh and C. A. Anderson, Am. J. Epidemiol.,
2010, 172, 1045–1052.
127 T. W. D. Edwards and D. A. Walker, C3, C4: Mechanisms,
and
Cellular
and
Environmental
Regulation,
of
Photosynthesis, Blackwell Scientic Publications, Oxford,
1983.
128 T. Whelan and W. M. Sackett, Plant Physiol., 1973, 51, 1051–
1054.
129 P. H. Raven, R. F. Evert and S. E. Eichhorn, Biology of Plants,
Worth Publishers, New York, NY, 5th edn, 1992.
814 | J. Anal. At. Spectrom., 2014, 29, 795–816
Tutorial Review
130 J. S. Ruff, A. K. Suchy, S. A. Hugentobler, M. M. Sosa,
B. L. Schwartz, L. C. Morrison, S. H. Gieng,
M. K. Shigenaga and W. K. Potts, Nat. Commun., 2013, 4,
2245.
131 M. Heasman, Br. Food J., 1989, 91, 9–16.
132 R. B. O. Ervin and C. L. Ogden, NCHS Data Brief, 2013,
article no. 122.
133 in Economic Research Service Yearbook Tables, USDA, 2013.
134 V. S. Malik, A. Pan, W. C. Willett and F. B. Hu, Am. J. Clin.
Nutr., 2013, 98, 1084–1102.
135 L. Chen, L. J. Appel, C. Loria, P. H. Lin, C. M. Champagne,
P. J. Elmer, J. D. Ard, D. Mitchell, B. C. Batch, L. P. Svetkey
and B. Caballero, Am. J. Clin. Nutr., 2009, 89, 1299–1306.
136 E. A. Dennis, K. D. Flack and B. M. Davy, Eat. Behav., 2009,
10, 237–246.
137 B. M. Popkin, L. E. Armstrong, G. M. Bray, B. Caballero,
B. Frei and W. C. Willett, Am. J. Clin. Nutr., 2006, 83, 529–
542.
138 K. D. Brownell and T. R. Frieden, N. Engl. J. Med., 2009, 360,
1805–1808.
139 R. R. Briefel and C. L. Johnson, Annu. Rev. Nutr., 2004, 24,
401–431.
140 G. A. Bray, S. J. Nielsen and B. M. Popkin, Am. J. Clin. Nutr.,
2004, 79, 537–543.
141 R. S. Strauss and H. A. Pollack, JAMA, J. Am. Med. Assoc.,
2001, 286, 2845–2848.
142 C. L. Ogden, M. D. Carroll, B. K. Kit and K. M. Flegal, NCHS
Data Brief, 2012, 1–8.
143 D. A. Schoeller, J. Archaeol. Sci., 1999, 26, 667–673.
144 L. L. Tieszen, T. W. Boutton, K. G. Tesdahl and N. A. Slade,
Oecologia, 1983, 57, 32–37.
145 S. H. Nash, A. R. Kristal, A. Bersamin, S. E. Hopkins,
B. B. Boyer and D. M. O'Brien, J. Nutr., 2013, 143, 161–
165.
146 W. C. Willett, Nutritional epidemiology, Oxford University
Press, New York, NY, 2nd edn, 1998.
147 C. P. Dancey and J. Reidy, Statistics without maths for
psychology, Pearson Education Limited, Essex, England,
4th edn, 2007.
148 K. Nakamura, D. A. Schoeller, F. J. Winkler and
H. L. Schmidt, Biomed. Mass Spectrom., 1982, 9, 390–394.
149 D. A. Schoeller, M. Minagawa, R. Slater and I. R. Kaplan,
Ecol. Food Nutr., 1986, 18, 159–170.
150 L. Tieszen and T. Fagre, in Prehistoric Human Bone, ed. J.
Lambert and G. Grupe, Springer, Berlin Heidelberg, 1993,
pp. 121–155.
151 J. Yoshinaga, M. Minagawa, T. Suzuki, R. Ohtsuka,
T. Kawabe, T. Inaoka and T. Akimichi, Am. J. Phys.
Anthropol., 1996, 100, 23–34.
152 T. C. O'Connell and R. E. Hedges, Am. J. Phys. Anthropol.,
1999, 108, 409–425.
153 T. C. O'Connell, R. E. M. Hedges, M. A. Healey and
A. H. R. W. Simpson, J. Archaeol. Sci., 2001, 28, 1247–1255.
154 R. Bol and C. Pieger, Rapid Commun. Mass Spectrom., 2002,
16, 2195–2200.
155 K. J. Petzke, H. Boeing and C. C. Metges, Rapid Commun.
Mass Spectrom., 2005, 19, 1392–1400.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
Tutorial Review
156 K. A. Hatch, M. A. Crawford, A. W. Kunz, S. R. Thomsen,
D. L. Eggett, S. T. Nelson and B. L. Roeder, Rapid
Commun. Mass Spectrom., 2006, 20, 3367–3373.
157 A. M. Mekota, G. Grupe, S. Ufer and U. Cuntz, Rapid
Commun. Mass Spectrom., 2006, 20, 1604–1610.
158 A. H. Thompson, L. A. Chesson, D. W. Podlesak,
G. J. Bowen, T. E. Cerling and J. R. Ehleringer, Am. J.
Phys. Anthropol., 2010, 141, 440–451.
159 T. C. O'Connell, C. J. Kneale, N. Tasevska and G. G. Kuhnle,
Am. J. Phys. Anthropol., 2012, 149, 426–434.
160 L. O. Valenzuela, L. A. Chesson, G. J. Bowen, T. E. Cerling
and J. R. Ehleringer, PLoS One, 2012, 7, e34234.
161 J. G. Gragnani, M. E. P. E. Garavello, R. J. Silva,
G. B. Nardoto and L. A. Martinelli, J Hum Nutr Diet, 2013,
Epub ahead of print.
162 M. J. Babcock, J. Nutr., 1955, 55, 323–336.
163 D. Hewitt and R. W. Hillman, Am. J. Clin. Nutr., 1966, 19,
436–439.
164 A. BØYum, Scand. J. Immunol., 1976, 5, 9–15.
165 D. Shemin and D. Rittenberg, J. Biol. Chem., 1946, 166, 627–
636.
166 S. H. Nash, A. Bersamin, A. R. Kristal, S. E. Hopkins,
R. S. Church, R. L. Pasker, B. R. Luick, G. V. Mohatt,
B. B. Boyer and D. M. O'Brien, J. Nutr., 2012, 142, 84–
90.
167 M. T. Sullivan, R. Cotten, E. J. Read and E. L. Wallace,
Transfusion, 2007, 47, 385–394.
168 V. E. Hedrick, D. L. Comber, P. A. Estabrooks, J. Savla and
B. M. Davy, J. Am. Diet. Assoc., 2010, 110, 1227–1232.
169 W. L. Clarke, D. Cox, L. A. Gonder-Frederick, W. Carter and
S. L. Pohl, Diabetes Care, 1987, 10, 622–628.
170 H. Elrick, L. Stimmler, C. J. Hlad, Jr and Y. Arai, J. Clin.
Endocrinol. Metab., 1964, 24, 1076–1082.
171 C. Fromentin, D. Tomé, F. Nau, L. Flet, C. Luengo,
D. Azzout-Marniche, P. Sanders, G. Fromentin and
C. Gaudichon, Diabetes, 2013, 62, 1435–1442.
172 E. A. Dennis, A. L. Dengo, D. L. Comber, K. D. Flack, J. Savla,
K. P. Davy and B. M. Davy, Obesity, 2010, 18, 300–307.
173 M. Konig, S. Bulik and H. G. Holzhutter, PLoS Comput. Biol.,
2012, 8, e1002577.
174 F. Q. Nuttall, A. Ngo and M. C. Gannon, Diabetes/Metab.
Res. Rev., 2008, 24, 438–458.
175 M. A. Permutt, J. Wasson and N. Cox, J. Clin. Invest., 2005,
115, 1431–1439.
176 P. Felig, T. Pozefsk, E. Marlis and G. F. Cahill, Science, 1970,
167, 1003–1004.
177 G. Perriello, R. Jorde, N. Nurjhan, M. Stumvoll, G. Dailey,
T. Jenssen, D. M. Bier and J. E. Gerich, Am. J. Physiol.,
1995, 269, E443–E450.
178 C. Waterhouse and J. Keilson, J. Lab. Clin. Med., 1978, 92,
803–812.
179 R. D. Yang, D. E. Matthews, D. M. Bier, Z. M. Wen and
V. R. Young, Am. J. Physiol., 1986, 250, E39–E46.
180 L. J. Reitsema, Am. J. Hum. Biol., 2013, 25, 445–456.
181 V. L. Fulgoni, 3rd, Am. J. Clin. Nutr., 2008, 87, 1554S–1557S.
182 L. A. Trivelli, H. M. Ranney and H. T. Lai, N. Engl. J. Med.,
1971, 284, 353–357.
This journal is © The Royal Society of Chemistry 2014
JAAS
183 H. F. Bunn, D. N. Haney, S. Kamin, K. H. Gabbay and
P. M. Gallop, J. Clin. Invest., 1976, 57, 1652–1659.
184 C. Weykamp, W. G. John and A. Mosca, J. Diabetes Sci.
Technol., 2009, 3, 439–445.
185 R. Hanas, G. John and A. C. C. International Hb, Clin.
Chem., 2010, 56, 1362–1364.
186 S. Ferri, S. Kim, W. Tsugawa and K. Sode, J. Diabetes Sci.
Technol., 2009, 3, 585–592.
187 L. Liu, S. Hood, Y. Wang, R. Bezverkov, C. Dou, A. Datta and
C. Yuan, Clin. Biochem., 2008, 41, 576–583.
188 J. Lee, M. Kim, H. Chae, Y. Kim, H. I. Park, Y. Kim, H. Chi
and H. J. Kwon, Clin. Chem. Lab. Med., 2013, 1–8.
189 in National Agriculture Statistics Service: Crop Production,
USDA, 2013.
190 M. Schwertl, K. Auerswald, R. Schäufele and H. Schnyder,
Agric., Ecosyst. Environ., 2005, 109, 153–165.
191 C. Gillam, Sweet times for cows as gummy worms replace
costly corn feed, ed. L. Gevirtz, Reuters, Reuters U.S.
Edition, 2012, vol. 2013.
192 O. Schmidt, J. M. Quilter, B. Bahar, A. P. Moloney,
C. M. Scrimgeour, I. S. Begley and F. J. Monahan, Food
Chem., 2005, 91, 545–549.
193 B. Bahar, F. J. Monahan, A. P. Moloney, P. O'Kiely,
C. M. Scrimgeour and O. Schmidt, Rapid Commun. Mass
Spectrom., 2005, 19, 1937–1942.
194 M. Boner and H. Forstel, Anal. Bioanal. Chem., 2004, 378,
301–310.
195 E. Piasentier, R. Valusso, F. Camin and G. Versini, Meat Sci.,
2003, 64, 239–247.
196 A. H. Jahren and R. A. Kra, Proc. Natl. Acad. Sci. U. S. A.,
2008, 105, 17855–17860.
197 A. H. Jahren and B. A. Schubert, Proc. Natl. Acad. Sci. U. S. A.,
2010, 107, 2099–2101.
198 R. J. Panetta and A. H. Jahren, Rapid Commun. Mass
Spectrom., 2011, 25, 1373–1381.
199 A. Barrie, S. T. Brookes, S. J. Prosser and S. Debney, Fert.
Res., 1995, 42, 43–59.
200 N. J. P. Owens and A. P. Rees, Analyst, 1989, 114, 1655–1657.
201 M. Minagawa and E. Wada, Geochim. Cosmochim. Acta,
1984, 48, 1135–1140.
202 T. S. Adams and R. W. Sterner, Limnol. Oceanogr., 2000, 45,
601–607.
203 S. Ambrose and M. DeNiro, Oecologia, 1986, 69, 395–406.
204 M. P. Richards and R. E. M. Hedges, J. Archaeol. Sci., 1999,
26, 717–722.
205 S. H. Nash, A. R. Kristal, A. Bersamin, K. Choy,
S. E. Hopkins, K. L. Stanhope, P. J. Havel, B. B. Boyer and
D. M. O'Brien, Eur. J. Clin. Nutr., 2014, 68, 91–96.
206 D. M. O'Brien, A. R. Kristal, M. A. Jeannet, M. J. Wilkinson,
A. Bersamin and B. Luick, Am. J. Clin. Nutr., 2009, 89, 913–
919.
207 F. Huelsemann, K. Koehler, H. Braun, W. Schaenzer and
U. Flenker, Am. J. Phys. Anthropol., 2013, 152, 58–66.
208 M. L. Disis and J. T. Slattery, Sci. Transl. Med., 2010, 2,
22cm29.
209 L. M. Bennett and H. Gadlin, J. Invest. Med., 2012, 60, 768–
775.
J. Anal. At. Spectrom., 2014, 29, 795–816 | 815
View Article Online
Open Access Article. Published on 10 February 2014. Downloaded on 16/04/2015 23:08:01.
This article is licensed under a Creative Commons Attribution-NonCommercial 3.0 Unported Licence.
JAAS
210 R. J. Amdur and C. Biddle, JAMA, J. Am. Med. Assoc., 1997,
277, 909–914.
211 Br. J. Anaesth., ed. C. Reilly, 2011, 107, 116–117.
212 in 111.4b, ed. J. H. S. o. Medicine, JMS Office of
Human Subjects Research-Institutional Review Boards,
2006.
816 | J. Anal. At. Spectrom., 2014, 29, 795–816
Tutorial Review
213 A. N. Westover and L. B. Marangell, Depression Anxiety,
2002, 16, 118–120.
214 S. Seely and D. F. Horrobin, Med. Hypotheses, 1983, 11, 319–
327.
215 G. Hems, Br. J. Cancer, 1978, 37, 974–982.
216 M. Virkkunen, Nutr. Rev., 1986, 44, 94–103.
This journal is © The Royal Society of Chemistry 2014